Fabrication and characterization of nanocomposite-based ... · mirrors [6,7,8], reﬂective thin...

Scuola di Dottorato in Fisica, Astrofisica e Fisica Applicata

Dipartimento di Fisica

Corso di Dottorato in Fisica, Astrofisica e Fisica Applicata

Ciclo XXVI

Fabrication and characterizationof nanocomposite-based

elastomeric optical devices

Settore Scientifico Disciplinare FIS/03

Supervisor: Professor Paolo MILANI

Coordinator: Professor Marco BERSANELLI

PhD Candidate:

Cristian Ghisleri

Academic Year 2013/2014

Commission of the final examination:

External Referee:Herbert Shea

External Member:Giovanni Mattei

Internal Member:Paolo Milani

Final examination:

21 January 2014

Universita degli Studi di Milano, Dipartimento di Fisica, Milano, Italy

To my parents, to Gabriele

MIUR subjects:

FIS/03

PACS:

78.67.Sc42.79.-e78.67.Bf

Contents

Introduction vii

I Deformable optical devices and nanocomposites 1

1 Adaptive and Tunable Optics 31.1 Deformable optical devices for adaptive and tunable optics 5

1.1.1 Deformable mirros 51.1.2 Tunable gratings 6

1.2 Applications of Adaptive Optics 71.2.1 Astronomy 81.2.2 Ophthalmology 91.2.3 Microscopy 101.2.4 Photonics 11

2 Nanocomposite-based elastomeric optics 132.1 Elastomeric optics 13

2.1.1 Transparent elastomeric optics 132.1.2 Reflective elastomeric optics 14

2.2 Requirements for reflective elastomeric optics 152.2.1 Reflectivity 152.2.2 Surface morphology 16

2.3 Resilience and stretchability 172.4 Metallization of elastomeric optical devices 19

2.4.1 Liquid metals 192.4.2 Physical Vapor Deposition 20

2.5 Metal-elastomer nanocomposites 222.5.1 Chemical synthesis of nanocomposites 232.5.2 Ion Implantation 242.5.3 Supersonic Cluster Beam Implantation (SCBI) 26

v

vi Contents

3 Supersonic Cluster Beam Implanted nanocomposites 273.1 Supersonic Cluster Beam Deposition (SCBD) and SCBI 273.2 Working principle of Supersonic Cluster Beam Implantation 28

3.2.1 Supersonic Expansion 283.2.2 Implantation in soft polymers 303.2.3 Equivalent thickness and filler volume fraction 313.2.4 Molecular dynamics simulations of SCBI in PDMS 32

3.3 Supersonic Cluster Beam Implantation setup 333.3.1 Pulsed Microplasma Cluster Source (PMCS) 353.4.1 Size distribution of metal nanoparticles 37

3.4 Morphological and mechanical characterization of SCBI Ag/PDMS nanocom-posites 373.4.2 Penetration depth of metal nanoparticles 373.4.3 Elasticity of SCBI nanocomposites 383.4.4 Surface roughness 39

II Reflective stretchable optics 57

4 Fabrication and characterization of elastomeric optical devices 594.1 Fabrication of the devices 594.2 Morphological characterization 594.3 Optical properties 60

4.3.1 Reflectance 604.3.2 Spot size and linearity 61

4.4 Focusing properties 62

5 A simple scanning spectrometer based on a stretchable reflective grating 715.1 Grating fabrication 715.2 Optical setup 725.3 Calibration 735.4 Spectra acquisition 73

III Plasmonics 77

6 Surface Plasmon Resonance (SPR) 796.1 Theoretical models 80

6.1.1 Drude model 806.1.2 Mie theory 846.1.3 Maxwell-Garnett theory 866.1.4 Shell-core model 88

6.2 Bergman model 91

7 SPR in SCBI metal-polymer nanocomosites 957.1 Absorption with different equivalent thicknesses 957.2 Time evolution of the nanocomposite at RT 987.3 Thermal annealing 997.4 Effect of passivation on SPR 1027.5 Stretching measurements 103

Introduction vii

Conclusions and perspectives 108

Bibliography 109

List of Publications 122

Introduction

Motivation

Adaptive and tunable optics, consisting in the correction of perturbed light wavefrontsby means of deformable optical devices, was first introduced in the 50’s by Babcock [1]and developed in the 70’s by Buffington [2, 3] and Hardy [4] independentely. Nowadaysit represents a well established technology mainly exploited in astronomy in order to re-duce the effects of atmospheric perturbations limiting the attainable resolution of largeastrophysical telescopes [5]. Adaptive and tunable optics is exploited in ophthalmol-ogy, microscopy and photonics as well for the improvement of imaging performances.However exploitation of adaptive and tunable optics is nowadays limited to high-levelor prototype-level instrumentation because of the high complexity and fabrication costsof the deformable optical devices, responsible for the correction of aberrations in lightwavefronts. The fabrication of functional, simpler and cheaper deformable optical de-vices is fundamental in order to apply adaptive and tunable optics in more commonlyavailable instrumentation.

Such deformable optical devices typically consist in deformable mirrors or gratings,able to change their optical properties (shape, focal length or pitch) dynamically, ac-cording to the perturbations of the wavefront or the wanted outgoing optical features.The technologies on which such deformable optical devices rely (for example segmentedmirrors [6, 7, 8], reflective thin membrane [9], MEMS tunable gratings [10, 11]) suffer ofseveral drawbacks in terms of high weight and complexity or low deformability andtunability of the optical properties. New technologies are required, aiming to overcomethese issues.

The fabrication of non planar diffraction gratings, required in a number of opticalmounts for the correction of spherical or higher order aberrations affecting the diffractedbeam or to add focusing to diffractive capabilities, is of concern as well for the necessaryhigh fabrication costs. The possibility to easily fabricate low-cost arbitrarily shaped re-flective diffraction gratings would be a breakthrough for example in monochromoatorsor spectral imaging techniques.

Elastomeric optics represents a promising technology for the fabrication of opticaldevices on highly deformable and conformable elastomeric substrates [12]. Reflectiveelastomeric optics, obtained by metallization of elastomeric substrates, is particularlyinteresting because of the possibility to limit the dimensions of elastomeric device basedoptical instruments [13] respect to devices working in transmission. However the metal-lization of elastomeric substrates by classic coating techniques is problematic because of

ix

x Thesis overview

the low adhesion and resilience of the reflective metal layer deposited on the elastomersurface [14, 15, 16].

Metal-elastomer nanocomposites may represent an alternative and effective approachfor the fabrication of reflective elastomeric optical devices since the reflective metal isembedded in the elastomeric matrix and does not consist in a continuous rigid metallayer, solving adhesion and resilience problems of coating techniques. However cur-rently available nanocomposites synthesis techniques (chemical [17, 18] or metal ionimplantation [19, 20] approaches) do not guarantee the achievement of reflectivity andsurface smoothness required for deformable optical devices.

Driven by the motivations described above, the present thesis is devoted to the fab-rication, characterization and exploitation of metal-elastomer nanocomposite based de-formable optical components (mirrors and diffraction gratings) obtained by means ofSupersonic Cluster Beam Implantation (SCBI) [21]. SCBI allows implanting electricallyneutral metal nanoparticles (silver in this work) with low kinetic energy in elastomericsubstrates like Polydimethylsiloxane (PDMS). The optical and morphological character-ization of the Ag/PDMS nanocomposite will demonstrate that the issues encounteredfor the fabrication of reflective elastomeric optical components by currently availablenanocomposite synthesis techniques can be overcome by using SCBI.

In particular optical properties of reflective elastomeric optical components synthe-sized by SCBI are affected by Surface Plasmon Resonance (SPR) characterizing silverparticles of nanometric size implanted in the elastomeric matrix. A characterization ofSPR in light of the theoretical model describing the optical behavior of metal nanopar-ticles embedded in a dielectric matrix and upon applied strain is necessary for a betterunderstanding and control of the optical properties of the devices during the fabricationprocess.

Thesis overview

This thesis consists of three parts, for a total of seven chapters. The introductory Part I,composed by the first three chapters, describes the technology on which adaptive opticssystems are based and introduces nanocomposite-based elastomeric optics as a possi-ble solution for the replacement of existing deformable optical devices. This sectionends with the description of the Supersonic Cluster Beam Implantation (SCBI) techniqueused in this thesis work for the effective fabrication of the metal-polymer nanocompos-ite based stretchable optical devices, overcoming the issues typical of nanocompositesobtained by classical techniques. Part II, composed by chapter 4 and chapter 5, is dedi-cated to the description of the results obtained in the field of stretchable reflective optics.In part III, composed by chapter 6 and chapter 7, a theoretical overview of Surface Plas-mon Resonance and our preliminary experimental results on the uv-vis absorption ofmetal-polymer nanocomposites are presented.

Substantial fractions of this thesis appeared as refereed publications in scientific jour-nals and resulted in the filling of a patent.

Chapter 1. Adaptive and Tunable Optics: applications, working principles andcurrent techniques used in tunable and adaptive optics are presented, followed bya detailed analysis of the drawbacks characterizing the currently used deformableoptical devices.

Chapter 2. Nanocomposite-based elastomeric optics: elastomeric optics repre-sents a possible solution to the fabrication of optical devices, specifically designed

Introduction xi

to be deformable. However optical elements working in reflection require the met-allization of the elastomeric surface. Among different available metallization tech-nologies, nanocomposites represent the ideal solution, matching all the require-ments needed from both an optical and a mechanical point of view.

Chapter 3. Supersonic Cluster Beam Implanted nanocomposites: SupersonicCluster Beam Implantation (SCBI) of metal nanoparticles in elastomeric substratesis introduced in details, in light of the optical and mechanical requirements neededby elastomeric optical elements working in reflection. The state of the art on su-personic cluster beam implanted elastomer is reported in a paper published onAdvanced Materials [21], while the morphological characterization of the implantedelastomer in a paper published on Journal of Physics D: Applied Physics [22].

Chapter 4. Fabrication and characterization of elastomeric optical devices: thecharacterization of the optical properties of nanocomposite-based reflective stretch-able mirrors and diffraction gratings fabricated by SCBI is presented. In particularwe show that it is possible to apply the stretchable grating to non-optical gradecurved surfaces to add focusing properties to diffractive power. The results of thecharacterization are the subject of a paper published on Laser and Photonics Reviews([23]).

Chapter 5. A simple scanning spectrometer based on a stretchable reflective grat-ing: stretchable gratings can be used in actual optical mounts and the spectrum ofa dye acquired and compared with tabulated data and with spectra acquired withcommercial spectrophotometers. This chapter is based on a paper submitted toApplied Physics Letters ([24]).

Chapter 6. Surface Plasmon Resonance (SPR): metal nanoparticles are charac-terized by Surface Plasmon Resonance (SPR), theoretically described by variousmodels. A brief overview of these models is reported in this chapter.

Chapter 7. SPR in SCBI metal-polymer nanocomosites: the experimental evi-dences and analyses of the Surface Plasmon Resonance of the gold and silver im-planted nanocomposites undergoing different treatments are reported. Prelimi-nary SPR characterization of stretched nanocomposites is particularly interestingand promising for a better understanding of the nanoparticles dynamics and topol-ogy inside the polymer.

Part I

Deformable optical devices andnanocomposites

CHAPTER 1

Adaptive and Tunable Optics

Adaptive optics is a technology used for the improvement of the performance of opticalsystems by reducing the effect of wavefront distortions, first proposed in 1953 by Bab-cock [1] for astronomical applications. The first realization of a working adaptive opticssystem dates back to 1977, thanks to the availability and improvement of automatic pro-cessing power needed for the real-time elaboration of the wavefront shape and controlof the correction system, by Buffington [2, 3] and Hardy [4] independentely. From thenon adaptive optics was exploited in different other applicative fields requiring the cor-rection of wavefront distortions, in addition to astronomy.

The working principle of an adaptive optics system is depicted in Figure1.1a. Wave-front distortions can be corrected by using deformable optical devices (usually deformablemirrors), able to dynamically change their optical properties according to the perturba-tion of the incoming light. A wavefront sensor, typically a Shack-Hartmann mask con-sisting of an array of lenses (also called lenslets) of the same focal length, and a CCDcamera or a segmented photodiode [26], reads the perturbation of the incoming light.The local tilt of the wavefront across each lens can be then calculated from the displace-ment of the focal spot on the sensor respect to the expected position for an unperturbedwavefront. By sampling an array of lenslets, all of these tilts can be measured and thewhole wavefront distortion approximated (Figure1.1b). A computer elaborates the in-formation obtained by the wavefront detector and subsequently controls mechanical orelectrostatic actuators applying proper distortions to the deformable optical device. Theinteraction of the distorted wavefront with the deformed optical device results in anoutgoing planar unperturbed wavefront.

Deformable mirrors are typically used as corrective devices in adaptive optics, butother devices able to dynamically change their optical properties were also developed.Among them tunable diffraction gratings and lenses represent interesting devices. Themodification of the grating grooves spacing or height leads to a change in the diffractionresolution and wavelength diffracted at a given angle or in the efficiency of the diffrac-tion orders. Like the crystalline in eyes, the focal length of tunable lenses can be changedby applying deformations to the device that modify the curvature of the lens. In thesecases we talk about tunable, rather than adaptive, optics.

Deformable optical devices and their complex actuation mechanisms represent themain contribution to the high costs still characterizing adaptive optics systems. Thislimits the exploitation of adaptive and tunable optics to niche applications like large as-trophysical telescopes, ophthalmology, high-level laser systems and optical microscopy.The use of cheaper, more reliable devices and simpler actuation systems could representsa breakthrough in these fields, making adaptive and tunable optics commonly availableon wider markets.

3

4

Figure 1.1: (a) Schematics of an adaptive optics system where the aberrated wavefront is reflectedoff the deformable mirror resulting in a corrected diffraction limited wavefront passed on to adownstream camera. The beamsplitter takes a sample of the corrected wavefront to monitor thephase and send updates to the control electronics to close the loop to the deformable mirror. (from[25]). (b) Schematic working principle of a Shack-Hartmannn wavefront sensor. The displacementof the position of the perturbed wavefront focused by the lenslet array respect to the expectedposition on the photon sensor (typically a CCD camera) gives the information on the distortionaffecting the wavefront.

Adaptive and Tunable Optics 5

Figure 1.2: (a) The two most common approaches for the actuation of deformable mirrors: work-ing scheme of segmented mirrors (top) and reflective membranes (bottom) [26]. (b) The 10.4msegmented mirror of Gran Telescopio Canarias [28]. (c) MEMS based deformable mirror based ona continuous faceplate (OKO R© Technologies) [25].

1.1 Deformable optical devices for adaptive and tunable optics

A fundamental part of adaptive or tunable optics systems is represented by deformableoptical devices, typically consisting in deformable mirrors or diffraction gratings. Here anoverview of the main current technologies on which deformable mirrors or gratings arebased will be given.

1.1.1 Deformable mirros

Segmented mirrors, reflective thin metal membranes, and ferrofluids are the three main solu-tions for the fabrication of deformable mirrors for adaptive optics [27].

Segmented mirrors [6, 7, 8] represent the most commonly followed approach for itssimplicity and the possibility to exploit existent technologies for the fabrication of eachsegment. A segmented mirror consists in an array of small rigid hexagonal mirrors eachactuated by hydraulic or electric pistons for their orientation according to the perturba-tion of the incoming wavefront (see Figure1.2a for the working scheme and Figure1.2bfor an example of a large segmented mirror). The number of segments can vary fromfew units to few tens and its increase requires an exponential increase of the computingpower needed for the actuation. The maximum number of segments for a single seg-mented mirror was achieved by Hulburd in 1989 [29] with an array of 512 mirrors andover than 10000 constituting elements. Despite their simple concept, segmented mirrorssuffer of some drawbacks: they require a high number of expensive actuators and thesoftware needed for the reading of the wavefront and their control is complex. Moreoverthe high weight of the elements needed for the mirror actuation requires robust support-ing structures. Additionally the gap between the segments is of concern because of lightlosses, extra infrared thermal emission and diffraction of light [27]. This last issue is ofparticular importance when very high resolutions must be achieved, as in the case ofophthalmology (blood capillaries or single cone cells) [30] and it can be overcome by theuse of continuous faceplate mirrors.

Continuous faceplate mirrors are typically based on reflective membranes (e.g. alu-minized polymeric membranes) whose shape is modified by push-pull actuators basedon piezoelectric materials [9] (Figure1.2a). The spatial continuous deformation of themembrane and the absence of gaps, typical of segmented mirrors, represent the main ad-

6 1.1 Deformable optical devices for adaptive and tunable optics

Figure 1.3: (a) Ferrofluids without (left) and with (right) an applied external magnetic field: con-trolling the magnetic field the mirror can assume the wanted shape (images from University ofLaval). (b) Actuating mechanism of a ferrofluid based mirror: the magnetic field is modulated bya series of coils [33].

vantages of this class of deformable mirrors. However very thin membranes are requiredfor proper shape modification and wavefront correction, making these deformable mir-rors highly fragile [27]. Thicker reflective membranes can be used and bent for the cor-rection of low-order aberrations. MEMS based continuous faceplate mirrors are alreadyavailable on the market (Figure1.2c [25]) at a very high cost.

Novel techniques are emerging in the field of deformable mirrors: among them theuse of liquid mirrors is particularly attractive in terms of simplicity and fabrication costs.Two approaches are followed: in the first a simple liquid metal like Mercury is used. Aperfect parabolic shape is assumed by the liquid metal contained in a vessel rotatingat a given angular speed. By changing the rotation speed the curvature, and thus thefocal length of the paraboloid, can be changed dynamically. A first telescope based ona 4m diameter liquid mirror is under construction in India (International Liquid MirrorTelescope, ILMT [31]). Despite the advantages in term of easy of fabrication and lowfabrication costs, issues concerning the toxicity and the weight of the liquid metals limittheir practical use.

Ferrofluids, whose shape can be modified by applying magnetic fields (Figure1.3)represent the second approach for the fabrication of liquid deformable mirrors. [32, 33,34, 35]. Ferrofluids consists in a ferromagnetic nanoparticles layer (size of approximately10nm) in suspension on a supporting liquid. Differently than liquid metals, ferrofluidbased mirrors can correct higher order aberrations with a precision dependant on thedimensions of the coils immersed in the liquid (Figure1.3). The difficulty of using bothliquid metals and ferrofluids based mirrors out of the horizontal plane represents themain limitation for the exploitation of such systems in many applicative fields.

From this brief overview on the field of adaptive optics, it is evident the need ofnew technologies for the fabrication of deformable mirrors, able to overcome the issuespresented above. In particular efforts are required in order to make adaptive opticssystems cheaper and available also for mass-market products. In certain applicativefields a particular concern consists in the dimensions of deformable mirrors (and thus tothe compactness of the instrumentation [30]). In this context MEMS mirrors seems to bethe way to follow toward miniaturization of optical apparatus [13] but such devices arestill highly expensive.

1.1.2 Tunable gratings

Tunable gratings are currently based on MEMS micromachined on silicon substrates.The first attempt to fabricate a deformable grating modulator dates back to 1992 by Sol-gaard et al. [10]. The device consisted in a suspended silicon nitride/silicon membrane


Figure 1.4: Scanning electron microscopy (SEM) picture of a fabricated micromechanical pitch-tunable grating. In the central part of the image the grating beams [11].

and a voltage can be applied between the membrane and the substrate. When no voltageis applied the membrane behaves as a mirror, while when the voltage is turned on themembrane is attracted by the substrate because of the electrostatic force and a depressionis created; if depressions are periodic, the three dimensional structure of a grating arises.It is thus possible to switch from mirror to grating behaviors by applying voltages of theorder of few Volts, however the control of the efficiency and of the pitch of the grating isnot possible.

Recently the control of the grating pitch in MEMS based devices was added, for ex-ample, by Yu et al. in 2010 [11]. The grating consists in a sequence of connected siliconbars with dimensions 500µm x 10µm and 5µm spaced. The bars are anchored to twocomb drives on the two opposite sides, able to increase the spacing between the barsand thus to change the angle toward which the diffracted beam is directed. This pitch-tunable diffraction grating can work both in reflection and in transmission since the op-tically active area of the device is free standing. The tunability is quite limited (less than4%) due to the stiffness of the silicon connection between the bars.

The well-established fabrication process, compatible with micromachining used inmicroelectronics, and the limited size of the devices, ideal for the fabrication of com-pact optical instrumentation, represent the main advantages of MEMS tunable gratings.Nevertheless the range of tunability is highly limited by the stiffness of the materialsused, not designed to be deformable, and to the poor movement capability proper of theactuation mechanisms currently employed.

1.2 Applications of Adaptive Optics

As already discussed at the beginning of this chapter, adaptive and tunable optics arecurrently exploited in niche applicative fields. Here an overview of the applications of

8 1.2 Applications of Adaptive Optics

adaptive and tunable optics in astronomy, ophthalmology, photonics and microscopywill be given. Disadvantages, open issues and requirements still not achieved by de-formable optical devices will be highlighted.

1.2.1 Astronomy

Adaptive optics was first introduced in Astronomy [1] in order to increase the resolv-ing power attainable with optical telescopes on Earth. Since their invention by GalileoGalilei in 1609 and the development of theory of optics by Isaac Newton in 1704, theincrease of the resolving power of optical telescopes has been a driving force in the de-velopment and design of new instruments. As known from basic optical theory, theresolving power R of a telescope is related to the wavelength λ of the light observedand the aperture diameter D of the instrument, as described by equation 1.1 for circularapertures:

R = 1.22λ

D(1.1)

Since the wavelength is fixed, attempts to build telescopes with larger apertures duringthe last three centuries have been made. The 102cm refractive telescope placed at theYerkes Observatory in Wisconsis, USA (1897) represents the largest refractive telescopeeven built, but larger aperture diameters can be achieved following a mirror-based ap-proach. In this direction it became soon evident that the attainable theoretical resolution,due to the diameter of the aperture of the tube, is strongly limited by atmospheric turbu-lence, even in the best sky conditions.

Assuming the atmosphere as steady and optically neutral, the planar wavefrontscoming from outer space would not be affected and could be focalized in just one singlepoint by an ideal optical focusing device, like an infinitely large lens or mirror. HoweverEarth atmosphere is not homogeneous: its density, temperature and composition varycontinuosly on a millisecond time scale. As a consequence, the wavefront coming fromspace is perturbed accordingly and therefore is not focalized in a single point anymore.The angular dimension of the light spot caused by atmospheric turbulence representsthe effective resolution attainable. This phenomenon limits the details observable in as-tronomical objects and makes the construction of even larger telescopes unuseful.

The problem of atmospheric turbulence (also referred as seeing) in astronomical ob-servations can be overcome by eliminating the atmosphere and sending telescopes inspace. This solution gathers the advantages of eliminating the problem of seeing andincreasing the wavelength window observable by instruments. With this aim, HubbleSpace Telescope (HST) was designed, built and launched, and in the last twenty yearsprovided us images unattainable with optical telescopes on Earth. Space telescopes arean effective solution but they are highly costly: the construction and launch of HST re-quired an expense of approximately 2.5 billion dollars, while its successor, James WebbSpace Telescope, estimated cost should be around 6.5 billion dollars [36].

An alternative cheaper solution consists in Adaptive Optics: the theoretical resolvingpower of optical instruments can be achieved by removing the effects of atmosphericturbulence from the wavefronts arriving on Earth by using deformable optical devices,according to the scheme presented in Figure1.1. Atmospheric perturbations typically oc-cur on a millisecond timescale, thus deformable mirrors must be actuated at frequenciesranging between few tens and few hundreds Hertz [27].

Nowadays adaptive optics is exploited in all the largest astronomical observatoriesof the world [5] as an effective solution for the problem of seeing. This technology led tothe design and construction of huge telescopes, up to 10.4m of diameter (Gran Telescopio


Figure 1.5: (a) Neptune observed with Keck II telescope equipped with adaptive optics. (b) Thesame image of Neptune without the use of adaptive optics. (from [5]).

Canarias, GTC) with much lower required costs than space telescopes (130 million eurosneeded for GTC [28]). The difference between an image acquired by Keck II telescopewith and without adaptive optics is shown in Figure1.5.

1.2.2 Ophthalmology

Spectacles for correcting eyes aberrations are used since thirteenth century but only inthe nineteenth century a full theoretical study of defects like defocusing and astigmatismin eye vision were given by Young [37]. Higher order aberrations, referred as irregularastigmatism, were later discovered by Helmoltz [38]. Such high-order aberrations arecaused by imperfections in the retina and can not be corrected by simple cylindricallenses. Retinal imaging is necessary for the detection of irregular astigmatism: the aber-rated wavefront of light reflected by the retina is mapped [39] by measuring the phaseshift of light coming from different points of the retina. In the last two decades the tech-nology on which retinal imaging is based underwent a rapid improvement thanks to theapplication of adaptive optics in ophthalmoscopy [40] and of the Shack-Hartmann masksensor for the detection of the perturbed wavefront coming from the eye [41, 42]. Retinalimaging became an easy task requiring only some milliseconds for a complete scan [26],thanks to the development of faster computing systems able to quickly elaborate hugeamounts of data. These progresses opened the way to closed-loop adaptive optics sys-tems for the eye and enabling the measurement of the time-dependant microfluctuationsin the eye’s wave aberration [43].

Closed-loop adaptive optics in ophthalmology is nowadays largely used both in ther-apy and diagnostics. Retinal imaging can be highly improved: the wavefront reflectedby the retina can be corrected and aberrations up to the 10th order can be detectedthrough the analysis of the corrections imposed by the adaptive optics system [42]. Thisrepresents a breakthrough in wavefront-guided refractive surgery for the permanent im-provement of eye’s vision of a patient. For example an excimer laser, treating the affectedarea of the retina and correcting defocusing, astigmatism and high-order aberrations, canbe controlled by the real-time measure of the retina-reflected wave aberration.

Adaptive optics applied to ophthalmology can serve also for the diagnosis of a num-ber of diseases (see Figure1.6). For example accurate measurements of the blood vessels


Figure 1.6: (a) 1 degree images of the same retinal location at two different focal planes in the righteye of a living human subject taken by exploiting adaptive optics. In the left image, capillaries assmall as 6µm are resolved. By focusing deeper into the retina, the underlying photoreceptors areresolved and the capillaries appear as faint shadows. (b) The left slide shows a sum of 5 imagesof a patient with retinal dystrophy. The photoreceptor array is patchy compared with a set of 6images from a healthy eye (from [30]).

structure and functionality are important for the detection and monitoring of some reti-nal diseases like diabetic retinopathy, and are commonly done with injected contrastagents or doppler methods, affected by a low lateral resolution and toxicity of the fluo-rescein agent [26]. Adaptive optics provides non-invasive and high resolution imagingof blood flow in vessels, avoiding the use of contrast agents [44].

The correction of eye’s aberrations also improves the imaging of the cone mosaic[30]), the study of the arrangement of S, L and M cones [45] and the detection of indi-vidual cone diseases. The high resolution attained with this technology will allow totest the effectiveness of treatment interventions and learning more about the mechanismof retinal diseases. Nevertheless some improvements are likely: deformable mirrors arethe most effective technology for adaptive optics, but they are usually characterized bylarge dimensions and high fabrication costs.

1.2.3 Microscopy

Imaging capabilities of optical microscopes are often affected by aberrations, compro-mising the resolution and the contrast achievable by the observer, particularly with highresolution instruments like confocal or two-photon microscopes [46]

The sample and the optical setup represent the two main causes for aberrations in mi-croscopy. In particular the large numerical aperture and the mismatch of the refractionindex between the lenses and the sample are the causes for the strong spherical aberra-tion often present in microscopes. These low-order aberrations can be easily correctedby applying field curvature correctors or by imaging with immersed objective in order tocreate a continuum between the refractive indexes of the lens and of the sample. Never-theless high-order or time-dependant aberrations are present, mainly when dealing withliving tissues, whose refractive index can be highly inohomogeneous in the vertical di-rection and in time. In order to overcome these problems, adaptive optics in microscopy


Figure 1.7: (a) Confocal microscopy inverted images of the focus spot on axis, for a right scanningof 70 mm and a down scanning of 50 mm (top left). The off-axis spots are displayed uncorrectedand corrected (top right and bottom). All the corrected spots and the centered one correspond toAiry patterns (from [47]). (b) Adaptive confocal fluorescence microscope images of labelled mouseintestine. Aberrations were determined using sequential modal wavefront sensing (from [48]).

was introduced [47], allowing for high resolution dynamic imaging of living tissues orreliable 3D scanning of semi-transparent biological samples. Figure1.7 shows the dif-ferences between the images obtained with a classic and an adaptive optics equippedconfocal microscope.

Moreover while confocal microscopy allows obtaining a 3D reconstruction of thesample, with two-photon microscopy only a bi-dimensional image can be acquired. Thisissue can be overcome with the use of deformable lenses, able to change their curvature,and thus their focal length, upon the application of an electric field or mechanical stress,literally squeezing the lens in the plane orthogonal to the optical axis [49]. Deformablelenses able to change their focal length in a range of up to 1mm, allows investigatingquite thick samples with the high resolution proper of two-photon microscopy.

1.2.4 Photonics

Photonics can take advantage of adaptive optics as well, in particular for the correctionof perturbed wavefronts inside laser cavities. Efforts in improving the intensity of exit-ing light have been concentrated on increasing the energy through the reduction of thepulses duration and their temporal coherence. Moreover wavefront distortions of thebeam are often neglected, even if they play a fundamental role in the beam intensity.Distortions in the wavefront affect the spatial quality and thus the beam ability to befocused, reducing the attainable focused intensity [50].

More interesting, tunable optics may allow to adjust the wavelength of a laser sys-tem. Usually laser cavities consists in a couple of mirrors continuously reflecting thelight passing through a gain medium, and are only able to provide monochromatic lightbeams, with a wavelength depending on the design of the optical cavity. The high in-tensity, collimation and coherence of laser light suggest its exploitation, for example,in gas sensing application, since gases are commonly characterized by a strong absorp-tion at given wavelengths ranging from 700nm to 3000nm [51]. Tunable lasers, typicallyDistributed Feedback lasers (DFB), represent an effective solution for such applications,avoiding the use of multiple monochromatic light sources. Mirrors of DFB lasers optical


cavity are replaced with tunable gratings able to select a given wavelength resonatingin the cavity by changing their pitch. Fabrication techniques and costs of such tunablegratings currently represent a limiting factor for the development and diffuse use of DFBlasers.

CHAPTER 2

Nanocomposite-based elastomeric optics

The widespread exploitation of adaptive and tunable optics system is mainly limitedby the high fabrication costs and technical issues of the deformable optical devices andby the complexity of their actuation systems as well. The main technological issues ofcurrently available deformable optical devices are related to the stiffness of the materialsused and the poor tunability and reliability of the devices.

2.1 Elastomeric optics

The introduction of the concept of Elastomeric Optics by Whitesides’ group in 1996 [12]represented an important breakthrough in the deformable optical elements fabricationtechnology. Elastomeric optical devices consist in optical components prepared usingelastomers, defined by IUPAC as polymers that display rubber-like elasticity [52], and de-signed to be reversibly deformable. A low Young modulus and a high failure strain incomparison with other materials make elastomers the ideal choice as substrates for thefabrication of deformable optical devices. Among them Polydimethylsiloxane (PDMS) isone of the most popular because of its low cost, nontoxicity and biocompatibility. PDMSis easy to process using standard microfabrication techniques, in particular replica mold-ing of existing masters on a scale suitable for optical applications (0.1-10µm size features)with high fidelity is possible, as it will be demonstrated in this thesis work.

Low fabrication costs, large tuning capabilities of the optical properties, durabilityand, in some instances, new types of performances generally characterize elastomericoptical devices. Such optical elements may not be comparable with existing rigid opticaldevices in terms of attainable optical quality, but the advantages due to their deformablenature overcome these possible limitations [12].

2.1.1 Transparent elastomeric optics

Elastomeric materials are usually highly transparent in the UV-vis range (300-800nmfor PDMS). This property makes elastomers the ideal choice for the easy fabrication oftransparent deformable optical devices such as lenses or diffraction gratings working intransmission. As already discussed in the previous section, deformable lenses can beexploited in those applications where a tuning of the focal length is needed, for examplein a 3D scan of biological samples with a two-photon microscope. Lenses are not the onlyobtainable transparent optical elastomer-based components: light valves [53], Fresnellenses [54], and transmission deformable gratings, both for scanning [55, 56, 57] (evenembedded in MEMS [58]) and sensing [59, 60] purposes, can be easily found in literature.

13

14 2.1 Elastomeric optics

2.1.2 Reflective elastomeric optics

Deformable optical elements such as mirrors or reflective diffraction gratings can be fab-ricated with elastomers, however metallization of the elastomer surface is required in orderto reach a sufficient reflectivity. Reflective elastomer-based optical devices potentiallyrepresent an effective solution for the drawbacks affecting deformable optical devicesneeded in adaptive optics, opening new opportunities for different technologies.

Elastomeric mirrors may represent a simpler, cheaper, more compact and reliable al-ternative to currently available deformable mirrors. The continuity of faceplate mirrors,the robustness of segmented mirrors together with a low weight, a high degree of de-formability and a low fabrication cost in a single elastomeric component may open theway to the exploitation of adaptive optics in a wider range of applicative fields. High-level amateur telescopes to novel optical components for satellites, where the weight ofeach single component is matter of concern, or the production of more compact oph-thalmoscopes [13] or portable and dynamic eye vision correcting instruments [61] rep-resent just few examples of applicative fields in which elastomeric mirrors can be ex-ploited. Moreover the high elasticity and conformability allow elastomeric mirrors tobe applied to arbitrarily shaped surfaces for the permanent correction of both low- andhigh-order aberrations as well. In particular the problems of astigmatism, spherical aber-ration and coma caused by off-axis alignment of optical components, typical of severaloptical mounts [62, 63] can be solved by applying elastomeric mirrors to properly de-signed curved surfaces. In addition elastomeric optical devices can be characterized byan extremely smooth surface, as it will demonstrated in Section 3.4.4. This representsa main breakthrough in the optical instrumentation fabrication process since an opti-cal grade supporting curved surface is not required, considerately reducing the devicemachining cost and complexity respect to the state of the art.

Similarly to elastomeric mirrors, reflective elastomeric gratings could represent abreakthrough in the downscaling of optical analytical instrumentation in terms of di-mensions and price as well. Reflective diffraction gratings are much more commonlyused in commercial optical mounts than transmission grating systems: the dimensionof optical systems can be reduced by folding the optical paths using reflecting devices.Moreover, reflective gratings are not limited by the transmission properties of the trans-parent substrate and can operate at much higher diffraction angles [63].

Elastomeric reflective diffraction gratings [12] represent the typical example of elas-tomeric optical devices. In classic rigid gratings, currently available on the market, thediffracted wavelength λ at a given angle θo is selected by rotating the grating. The changeof the incidence angle θi of the impinging light selects the wavelength diffracted at theangle θo, according to the well-known law of planar gratings:

λ = dsin θo − sin θi

m(2.1)

where d is the pitch of the grating and m is the diffraction order considered. Thisapproach, commonly used in most of monochromators and spectrometers, requires ahighly precise rotation mechanism, increasing the cost of the optical instrumentation.Elastomeric gratings allow to keep the angle of incidence of the impinging light θi fixed.The diffracted wavelength λ at a given angle θo is selected exploiting the change of thegrating pitch dwhen the device is stretched, as shown in Figure2.1, requiring cheaper andeasier mechanisms.

The most appealing property of these devices consists in the possibility to easily andconveniently add optical power to diffractive elements by applying elastomeric gratings

Nanocomposite-based elastomeric optics 15

Figure 2.1: Strain applied to an elastomeric diffraction grating changes its grooves periodicity andalters the spacing of the diffraction pattern [12].

to arbitrarily shaped non-optical grade surfaces, as in the case of elastomeric mirrors.The limited number of focusing and correcting components needed in optical mountsand the aberration-free imaging capability of such devices would open the way to thedesign, for example, of cheaper, smaller, and portable spectrographs available on a widermarket and applicable to a wider range of fields.

Despite the clear advantages elastomeric optical elements working in reflection wouldbring in the design of optical mounts, this technology was poorly developed since theirfirst appearance in 1996 [12]. The lack of an effective technique for the metallizationof the elastomeric surfaces meeting the strict requirements needed for reliable reflectiveelastomeric optical devices can be considered the main cause for this lack of develop-ment of the technology.

2.2 Requirements for reflective elastomeric optics

2.2.1 Reflectivity

Reflective optical devices are usually designed to work in a given wavelength rangeaccording to the application in which are supposed to be exploited. Commercially avail-able mirrors or gratings are usually characterized by a reflectivity between 90% and100% and by a uniform reflectance spectrum. The same performances in terms of reflec-tivity are not strictly required by deformable optical devices since their high deformabil-ity represents the real added value of such devices [12]. Although no reference data arepresent in literature for the reflectivity required in elastomeric optical devices, we can as-sume that a reflectivity greater than 60% should be sufficient for most of the applicationsmentioned in the previous chapter.

16 2.2 Requirements for reflective elastomeric optics

Figure 2.2: Scattered light patterns vs. surface roughness. (a) Light scattering pattern vs. rough-ness. (b) light speckle pattern vs. roughness. Here Ra represents the arithmetic average of absolutevalues roughness, defined as Ra = 1

n

∑ni=1 |yi| (from [64]).

2.2.2 Surface morphology

Surface smoothness is a fundamental property of each optical component, working bothin reflection and in transmission, and is quantified by the surface roughness Rq (in thiscontext only the root mean squared roughness will be considered) defined as:

Rq =

√√√√ 1

n

n∑i=1

y2i (2.2)

where yi represents the deviation of the measured vertical profile of the surface fromthe mean vertical level and n the number of points measured for the real profile. Sur-face roughness is responsible for the scattering of the incident light by the surface of theoptical device, as can be seen in Figure2.2.

Scattering consists in a random diffusion of the impinging light intensity by a roughsurface. Surface roughness must be therefore finely controlled and minimized. In 1961Bennett and Porteus came up with a theoretical model describing the relationship be-tween the reflectance of a surface and its RMS roughness [65]. Beckmann and Spizzichinoin 1965 introduced the concept of Total Integrated Scatter (TIS, the total power of lightscattered into the hemisphere above the surface divided by the power incident uponthe surface) in optics [66] and found an analytical relationship between TIS and surfaceroughness:

TIS (Rq) = 1− R

R0= 1− e−

(4πRq cos θi

λ

)2

(2.3)

where R is the reflectance of the scattering surface, R0 is the reflectance of the ideallyflat surface, θi is the angle of incidence respect to the normal of the surface and λ thewavelength of light. As can be seen from equation 2.3, the fraction of scattered lightdepends on the surface roughness, as expected, and on the angle of incidence of light aswell. Figure2.3 shows the trend of the TIS function as a function of both RMS roughnessRq and the incidence angle θi. Optical devices working in reflection should be charac-terized by a TIS not larger than 5% in order to achieve a good optical quality. Extremelylow RMS roughnesses are thus required, especially for light impinging normally or withlow angles of incidence to the surface (Figure2.3a). Moreover, deformable optical de-vices applied to highly curved surfaces and used in off-axis configurations requires a


Figure 2.3: (a) Trend of the TIS function (fraction of light scattered by a rough surface) as a functionof the surface roughness for different angles of incidence. (b) Trend of the TIS function as a functionof the incident angle for different surface rougnesses.

uniform and limited (below 5%) scattering percentage for a wide range of light incidentangles. This requirement is again guaranteed by an extremely low surface roughness.As reported in Figure2.3b, the scattering relative to a roughness of 10nm is below 6%(upper limit with normal incident light) and is constantly below 5% for incident angleslarger or equal to 26◦.

2.3 Resilience and stretchability

Reflectivity and surface smoothness are fundamental requirements easily satisfied byclassic coating techniques currently exploited for the metallization of reflective opticalelements. However the optical response of reflective elastomeric optical componentsstrongly depends on additional properties required for such devices and related to re-flectivity and surface morphology: resilience and stretchability.

Elastomeric optical devices are designed to be deformable, thus to change their shapeand/or dimensions. Each deformation, even simple bending, can be led back to a stretch-ing of the material, strictly related to the adhesion of the rigid reflective metal layers tothe polymer surface by means of the Young modulus mismatch between the two ma-terials. Generally when a thin metal film on a polymer substrate is stretched, the rigidmetal film is not able to follow the deformations of the underlying elastomer and en-counters rupture and delamination at strains ranging from few percents up to few tenspercents in best cases [67]. This range of variability of the rupture threshold depends onthe adhesion of the metal thin film to the polymeric substrate. In the case of no adhesion,the maximum strain achievable by a freestanding thin metal film without encounteringbreakage is less than 1-2% [68, 69, 70]. The delamination mechanism of the metallic filmwhen mechanically stressed is schematically shown in Figure2.4. When a strain is ap-plied dislocations in the metal lattice arise leading to the formation of necks on the metallayer [68, 71, 67], where all the stress caused by the applied strain concentrates. The neckgets thinner and thinner yielding to the rupture of the metal layer where the neck arose.

18 2.3 Resilience and stretchability

Figure 2.4: (a) Free-standing thin metal film breakage takes place where the neck is formed, dueto the arising of dislocations in the metal lattice caused by the stress. (b) If the metal layer iswell adherent to the polymer surface, the stress on the metal layer is dissipated by the polymer,increasing the stretchability of the former (image from [67]).

If the metal film is well adherent to the polymer surface, the stress accumulated on theneck is partially dissipated by the subtending polymer (see Figure2.4b), increasing thebreakage threshold. However the formation of the neck gives rise to a point in whichthe metal film is detached from the polymer surface and here delamination takes place.The control on the formation of microcracks in metal layers on top of polymeric surfaces[14, 72] can be useful in order to limit and improve the stretchability of such systems[73, 74]. However microcracking and delamination dramatically increase the surfaceroughness of the metallized polymer, thus affecting the optical quality of the device.An optimal adhesion of the metal reflective layer is of fundamental importance in orderto avoid the formation of microcracks, preserving the optical quality of the elastomericreflective component when subjected to mechanical stress.

The breakage threshold of the rigid metal layer can be increased by improving itsadhesion with the elastomeric surface [71] thanks to deposition of a thin chromium ortitanium film between the polymer and the reflective coating [72, 75, 76, 77]. The betteraffinity of Cr or Ti with the elastomer helps in avoiding delamination, while crackingremains an issue since it causes a dramatic increase of the surface roughness on themicroscale, leading to scattering and other detrimental effects in the optical response ofthe device [78]. The mismatch between the Young moduli of the elastomeric surface andthe rigid metal layer deposited on it also causes the formation of buckles and wrinkleson the surface of the device [79, 80], independently on the adhesion between the twolayers. For example Guerrero et al. [81] observed the arising of unexpected diffractedbeams orthogonal to those due to the grooves of the grating. The formation of largeperiodic cracks or buckles in the stretching direction is supposed to be the cause forthe appearance of such secondary diffracted spots. Thinner metal layers show a betteradhesion as well, and larger maximum stretching thresholds (up to 32%) [82].

Stability under stretching is also important: the optical response must be reliable andstable in time and with the increase of the number of stretching cycles. A constant be-havior of the reflected or diffracted spots shape or angular dimension during a singleor after a certain number of stretching cycles is fundamental in order to ensure the op-tical quality of the device and to avoid continuous calibration procedures. Elastomericsubstrates and reflective layers must be then chosen accordingly.


Figure 2.5: Fabrication process of (a) elastomeric gratings and (b) elastomeric mirrors with liquidmetals [12].

2.4 Metallization of elastomeric optical devices

2.4.1 Liquid metals

The first attempt to fabricate reflective elastomeric mirrors or gratings dates back to 1996by Wilbur et al. [12]. The problems of cracking and delamination of reflective thin metalfilms on or in elastomeric substrates were overcome by the use of liquid metals (mercuryor gallium) embedded in the elastomer. The fabrication process is similar for reflectiveelastomeric gratings and mirrors as shown in Figure2.5.

The basic idea consists in replicating an existing rigid commercial grating or mirrorwith PDMS (in particular Dow Corning Sylgard 184) by molding technique, coveringthe optically active molded surface of the elastomer with mercury or gallium and finallyencapsulating the liquid metal by a second elastomeric layer. This technique completelyovercome the problems of metal adhesion and roughness of the reflective layer, how-ever liquid metals have a considerable weight (the densities of mercury and gallium are13.546g ·ml−1 and 5.907g ·ml−1 respectively) and a macroscopic thickness is needed toensure the complete coating of the surface, leading to deformations of the elastomericstructure. These deformations affect the optical response of the device, as can be seenin Figure2.6. The elastomeric structure of the device can be reinforced by increasing thePDMS thickness, however thicknesses larger than 2mm are detrimental for the opticalquality of the components, introducing a considerable light scattering, as stated by thesame authors [12].

In the fabrication process discussed above, molding is generally a fundamental stepfor the fabrication of elastomeric optical devices, however the metallization of the elas-

20 2.4 Metallization of elastomeric optical devices

Figure 2.6: Optical response of an elastomeric grating working in transmission (left) and in reflec-tion (b) with an encapsulated liquid metal (gallium) when subjected to stretching up to 10%. It isworth noting the slightly non-linearity of the metallized grating due to the deformations imposedby the weight of the liquid metal [12].

tomers surface for reflective purposes remains an open issue. In the following sections anoverview of the commonly available elastomer metallization techniques is given, with adetailed analysis of the main advantages and drawbacks.

2.4.2 Physical Vapor Deposition

One of the most common and widely used techniques in elastomeric optics is physicalvapor deposition (PVD) of metal atoms on the top of polymeric surfaces [82, 83, 81, 78].A liquid or solid metallic material is vaporized in atoms or molecules and transportedthrough a vacuum or low pressure gaseous environment, to the substrates where it con-denses. PVD is usually characterized by a high deposition rate (tenths of nanometersper second), allowing to deposit few nanometers up to few micrometers thick rigid andcontinuous metal films on the elastomer surface.

PVD techniques differentiate according to the method use for the evaporation of themetal:

• Thermal evaporation: Powders, pellets or wires of the metallic material are meltedin a tungsten or molybdenum crucible, whose evaporation temperature is muchhigher than the metal to deposit, and heated by Joule effect by means of the highcurrent (tens of amperes) flowing in the crucible (Figure2.7a). Evaporated metalatoms travel in vacuum toward the substrate to coat with an energy of approxi-mately 0.1eV, thus not causing any damage to the polymeric substrate. The heatgenerated by the crucible is of concern for the physical-chemical properties of theelastomer, especially for substrates with a low melting point (100-120◦C) and forlong deposition times.

• E-beam evaporation: An energetic electron beam (up to 15keV) is directed towardthe metal target that is locally heated beyond its evaporation temperature by thehigh energy of the electrons (Figure2.7b). Changing the electrons energy the de-position rate can be controlled (from tenths to some nanometers per second) whilethe evaporated metal atoms energy is comparable to thermal evaporation. The lo-cal heating of the metal target, and thus its reduced influence on the polymericsubstrate, represents the main advantage of this technique.


Figure 2.7: The three PVD processes: (a) Thermal evaporation, (b) E-beam evaporation, (c) Sput-tering.

• Sputtering: Ions are accelerated toward a metal target and the sufficiently highions energy (typically Ar+, with energies of some keV) extracts metal atoms fromthe target (Figure2.7c). Metal atoms reaching the elastomeric surface are character-ized by energies higher than the previous methods, between 1 and 100eV, leadingto a better adhesion to the polymeric substrate [84]. However the deposition rateis lower than thermal or e-beam evaporation.

The high deposition rate represents the main advantage of coating techniques likeevaporation, allowing reaching considerable thick metal layers on large areas in a rea-sonable amount of time, as required for the scalability of the fabrication process. Agood reflectivity (0.7-0.8) can be achieved [85] with reasonable thin metal layers, how-ever many are the drawbacks.

The stiffness of the continuous metal layer on the soft elastomer surface and the lowadhesion between the metal layer and the polymer are matter of concern because of itscracking and delamination when mechanically stressed [14, 72]. The maximum stretch-ing percentage after which cracking or plastic deformations of the metal layer take placeis limited to 7-8% of its original dimensions. Cracking arises even by using sputtering[75, 15, 14], involving higher impinging energies and so better adhesion between themetal and the polymer.

Condensation of metal atoms on the surface of elastomers and the subsequent forma-tion of a continuous metal layer induces a stress on the polymeric surface, leading to theformation of microwrinkles [79, 16] for deposition on prestretched substrates (in orderto improve the stretchability of the material) [76, 86] even without externally mechanicaldeformations. Despite the attempts to exploit these ordered spontaneous structures asstretchable optical devices (e.g. tunable diffraction gratings [87]), tunabilities larger than3.6% are not achieved. Moreover the control over the formation of microwrinkles is nottrivial and their dimension and periodicity may negatively affect the optical response ofthe evaporated elastomeric optical device.

In the end the high divergence and inhomogeneity of the atomic beam cause a strongshadowing effect, representing a limiting factor for the uniform coating of three-dimensionalstructures like gratings grooves. The highly irregular profile of an evaporated elas-tomeric grating caused by cracking, wrinkling and shadowing, will be presented in thenext chapter.

From this extensive discussion it is clear that rigid thin metal layers deposited on the

22 2.5 Metal-elastomer nanocomposites

Figure 2.8: (a) Optical microscopy images of wrinkles on a thermally evaporated PDMS surfacewith 5nm of Chromium and 50nm of Gold [79]. (b) SEM micrograph of evaporated PDMS surface.It is worth noting the formation of cracks even without mechanical solicitations applied to thesample. PDMS was e-beam evaporated with 5nm of Chromium and 25nm of Gold [72].

polymer surface do not represent an effective solution for the fabrication of reflectiveelastomeric optical devices. The obtained material does not match the optical qualityrequired for functional and reliable deformable optical device. Adhesion and delamina-tion can be improved by embedding a pre-formed metal layer inside the polymeric matrix,however cracking remains an issue.

2.5 Metal-elastomer nanocomposites

Nanocomposite materials can represent an effective solution for the fabrication of reli-able elastomeric optical devices overcoming the limitations of metal-polymer bi-layersin terms of resilience of the reflective layer [21]. A metal-polymer nanocomposite is acomposite material consisting of metallic particles with nanometric size embedded in apolymeric matrix. They were exploited in order to overcome the issues related to adhe-sion, cracking and delamination due to the presence of two layers with different stiffnessin metal-coated elastomers in different fields like stretchable electronics. In nanocompos-ites the metallic material consists in single metal nanoparticles able to slightly move faraway each other and re-arrange when a mechanical deformation is applied to the device,and to restore their original position when the material is released.

Despite the several advantages related to resilience and mechanical properties nano-composite materials could bring, no literature can be found in the field of nanocomposite-based elastomeric optics. This lack is due to the different synthesis techniques, notallowing to obtain nanocomposites characterized by appropriate optical properties interms of reflectivity and surface smoothness required for the fabrication of functionalelastomeric optical devices. In particular reflectivity and elasticity of metal-polymernanocomposites represent critical issues: in order to achieve reflectivities mentioned inthe requirements for elastomeric reflective optics, the polymer must be highly loadedwith metal nanoparticles. Such a high nanoparticle concentration is difficult to obtainwith nanocomposite synthesis techniques described below, and strongly affect the elas-ticity of the obtained material. The Young modulus of metal-elastomer nanocompositesincreases with the nanoparticles concentration (or filling factor) according to Guth equa-tion [88, 89], as shown in Figure2.9. A trade-off between a high loading of the polymericmatrix, needed in order to achieve a good reflectivity of the nanocomposite, and the elas-ticity of the obtained material, fundamental for deformable and stretchable applications,must be found.


Figure 2.9: Increase of nanocomposites Young modulus as a function of the concentration of themetal filler for different filler aspect ratios (α represents the ratio between the length and thebreadth of the filler particles) [89].

The main nanocomposites synthesis approaches are described in the following sec-tions.

2.5.1 Chemical synthesis of nanocomposites

Chemical synthesis represents the most common and simple technique for the forma-tion of metal-polymer nanocomposites and can be divided in two different approachesdepending on the way the nanoparticles are synthesized.

1. In-situ approach consists in the the simultaneous synthesis of metal nanoparticlesand the polymer by means of precipitation of polymer and nanoparticles startingfrom metal salts [17], thermal decomposition of metal precursors dissolved in thepolymer [90, 91], coevaporation or copsputtering of polymers and metal atomscondensating in metal nanoparticles [92, 93] or CVD and PVD with thermal treat-ments for particles aggregation [94].

2. In ex-situ methods nanoparticles are first prepared by a controlled precipitation of acolloidal precursor [90, 95, 96] or by other chemical or physical methods, and thenembedded in the polymer liquid monomer precursor by means of sol-gel methods[18, 97, 98, 99]. The mix of monomers and metal nanoparticles is then polymerized.Chemically synthesized nanoparticles often present an inert shell (usually metaloxide o functional groups) aiming to avoid clusters aggregation and deactivation[91].

In both cases a uniformly filled bulk nanocomposite is obtained.Three dimensional structures, like gratings grooves, can be easily obtained by mold-

ing technique of existing masters, the only requirement being the nanoparticles sizesmaller than the features to replicate. The surface roughness of the nanocomposite co-incides with the roughness of the elastomeric substrate (that in case of untreated PDMSis few nanometers) and depends on the polishing of the surface on which the mirror orgrating is molded. In such a way light scattering can be easily controlled by choosingthe appropriate polymer and by a proper machining of the molding master.


The granular nature of the reflective layer may also give rise to light scattering or ab-sorption, depending on the nanoparticles size. Light scattering can be limited by usingnanoparticles with a diameter d much lower than the light wavelength λ. Nanometricsized metal particles also present a strong absorption in the visible range due to Sur-face Plasmon Resonance (SPR) [100], whose wavelength peak depends on the size andconcentration of the nanoparticles.

The main drawbacks of the use of chemically synthesized nanocomposites consistin the use of chemicals for the production of metal nanoparticles and the poor reflec-tivity of the obtained nanocomposites. Many polymers, in particular elastomers, sufferof chemical-physical degradation when exposed to acids, bases or other chemicals, alsoused in the metal nanoparticles synthesis [101]. Some elastomers are also well known fortheir considerable swelling following solvent absorption [102]: this can alter the physicalproperties and affect the geometry of the elastomeric substrate.

Chemical synthesis (both-in situ or ex-situ) allows obtaining nanocomposites uni-formly filled with metal nanoparticles and, for in-situ approaches, the wanted clusterconcentration can be reached with a proper concentration of metal precursor. The pre-cursors must then be absorbed by the preformed elastomeric matrix, making thus verydifficult the achievement of filling factors larger than 30-50% [103]. If mixed with the liq-uid pre-polymer, the high precursor solvents concentration may impede the elastomerpolymerization. Ex-situ techniques present similar problems: the high concentrationof metal nanoparticles inside the liquid monomers solution avoids the monomer poly-merization because of the interaction between the polymer chains and the nanoparticlessurface [104, 105]. As a proof for the difficulty of obtaining highly loaded nanocom-posites by chemical synthesis, no works on elastomeric optical devices are reported inliterature.

In conclusion, the advantages of the nanocomposite approach in the fabrication ofelastomeric reflective devices remains however superior to classic coating techniques interms of stretchability and resilience of the reflective layer to the elastomeric substrate.A physical technique enabling the fabrication of metal-polymer nanocomposite charac-terized by higher particles concentrations, and thus reflectivities, is likely to be used.

2.5.2 Ion Implantation

Ion implantation, widely used in the microelectronic industry for the doping of siliconsubstrates, represents also an effective technique for the fabrication of metal-polymernanocomposites by means of implantation of metal ions or nanoparticles inside poly-meric matrices. The working principle of this technique is schematically represented inFigure2.10. An electric field accelerates metal ions or particles extracted from a sourceand ionized with different methods (electronic ionization, electric discharges, radioac-tive ion sources among them). Before a further acceleration and implantation in thesubstrate, a magnetic field deflects the ion beam toward a slit thanks to the Lorentz law.The exiting ions mass can be selected by tuning the intensity of the magnetic field. Theselected ions than undergo a second acceleration by a second electric field and get colli-mated by a series of magnetic quadrupoles. Finally two couples of scanning electrodesaddress the ions to the point of the polymer to implant.

The penetration depth of the ions in the polymer depends i) on the ions energy andii) on the polymer stopping power. The latter can be modeled by considering the processesinvolved in the ions energy loss during the interaction with the polymer. The two maincontributions to the ions energy loss are represented by the electronic stopping and thenuclear stopping. The dominant contribution depends on the acceleration voltage of the


Figure 2.10: Schematic of a typical Ion Implanter

ions. The main contribution for energies below 1-2MeV is due to the inelastic interactionof the ions with the atoms of the polymeric substrate (nuclear stopping), while for largerion energies, excitation and ionization of the atoms of the substrate take place, leading toelectronic stopping. Implantation in elastomers can be achieved for very low ion ener-gies, in the range between 2keV and 10keV [106] obtaining penetration depths betweenfew nanometers up to few tens of nanometers. In this regime nuclear stopping is themain responsible for the ions energy loss.

Ion implantation in elastomers is extensively studied and exploited for the fabrica-tion of electrodes for elastomeric actuators [107, 108, 57, 106]. A Filtered Cathodic Vac-uum Arc (FCVA) in which a metal target is vaporized by high-voltage pulses, producesa plasma containing metal ions, undesired nanoparticles and electrons. Metal ions areselected by a magnetic field and accelerated towards the polymeric substrate where theyget implanted with doses ranging between 0.9·1016 and 7·1016at/cm2 in order to reacha good conductivity [108]. It is worth noting that electrically charged atoms are im-planted in a polymeric substrate behaving like an insulator, leading to a polymer charg-ing, mainly in the first stages of the implantation process, and causing deformationsin the nanocomposite material. Polymer charging is limited by dissipation of electri-cal charges across the sample once the conductive percolation threshold is reached (ata minimum dose of 1.5 · 1015at/cm2 [106]). Metal ions inside the elastomer diffuse andaggregate, forming metal nanoparticles whose size can not be directly controlled.

Reflectivity in ion implanted materials is not widely studied and only few works[109, 110] consider the problem of ion implanted glasses. In these works reflectivities upto 40% are achieved with doses larger than 4 · 1016at/cm2, insufficient for the fabricationof reflective optical devices. Also Rosset et al. [19, 111] considered the optical propertiesof ion implanted PDMS for the fabbrication of arrays of tunable lenses, obtaining a trans-mittance not below 60% (and thus reflectivities not larger than 40%, as in the previouscase).

The main limitation for the achievement of higher reflectivities in ion implanted elas-tomers is the dramatic increase of the surface roughness following the implantation pro-cess. The nuclear interaction of ions with the polymer atoms during nuclear stoppingleads to sputtering, local stress modification, charging and heating of the polymer itself[20, 108]. PDMS surface is extremely smooth (RMS roughness below 2nm [112]) but in-creases up to approximately 100nm after Titanium or Palladium ion implantation at low


energies (5keV) [107]. As seen in the requirements needed for reflective elastomeric op-tical devices, roughness is a fundamental parameter that must be controlled in order toavoid light scattering. As shown in Figure2.3, the scattered light fraction ranges between50% and 60% for incident angles up to 30◦ in the case of a roughness of 40nm.

Furthermore a much larger increase of the nanocomposite stiffness (more than twoorders of magnitude) is observed respect to chemically synthesized nanocomposites andGuth’s theoretical expectations. The high energies involved in ion implantation, dissi-pated by the polymer nuclear stopping, cause the breakage and carbonization of the elas-tomeric chains. Carbonization consists in the breakage of bonds between hydrogen andcarbon of the PDMS methyl groups, leading to an increase of carbon atoms concentra-tion in the whole nanocomposite material. As a consequence, the chemical modificationof the polymeric chains leads to a deterioration of the physical and mechanical proper-ties of the polymer itself. This hypothesis is supported by the molecular theory of rubberelasticity [113] stating that the elasticity of a rubber-like polymer does not depend on thepolymeric chains interactions or arrangement, but only on the molecular composition ofthe chains.

From these considerations it is clear that ion implanted elastomers do not representoptimal materials for the fabrication of reflective optical devices and imaging purposes,as stated in [19], because of issues concerning both the optical quality of resulting mate-rials (reflectivity and light scattering) and mechanical properties.

2.5.3 Supersonic Cluster Beam Implantation (SCBI)

Recently Supersonic Cluster Beam Implantation (SCBI) of low energy electrically neutralmetal nanoparticles in soft polymers was developed by our group [114, 115] and ex-ploited for the effective fabrication of highly deformable and stretchable nanocomposite-based electrodes [21]. Successful results obtained in the field of stretchable electronicssuggest the exploitation of SCBI as effective nanocomposite synthesis technique for elas-tomeric optical devices as well.

As it will be extensively described in the next chapter, nanoparticles energies in-volved in this process ranges between tenths and few eV/atom, thus avoiding the poly-meric chains breakage occurring for ion implantation because of nuclear stopping. Asa result SCBI overcome the drawbacks of ion implanted elastomeric optical devicesin terms of polymer deterioration, surface roughness and thus optical quality of thenanocomposite surface, while achieving reflectivities required for elastomeric opticaldevices, not obtainable by chemical synthesis approaches. The increase of the Youngmodulus due to the loading of the elastomeric matrix typical of nanocomposites synthe-sized by chemical techniques is here overcome by concentrating the implanted nanopar-ticles in an extremely thin superficial layer, not considerably affecting the elasticity ofthe whole material.

CHAPTER 3

Supersonic Cluster Beam Implanted nanocomposites

3.1 Supersonic Cluster Beam Deposition (SCBD) and SCBI

SCBI derives from a technology developed in the last decade for the deposition of nanos-tructured thin films on rigid substrates and called Supersonic Cluster Beam Deposition(SCBD) [116]. The low nanoparticles kinetic energy and the substrate properties playa fundamental role in the production of porous nanostructured metal thin films. Ac-cording to molecular dynamics simulations [117] if the ratio between kinetic energy peratom of the metal cluster Eatk and the cohesive energy between the cluster and the sur-face Eclcoh is much lower than unity, the metal nanoparticles are not fragmented uponimpact with the rigid surface and the nanostructure preserved, as shown in Figure3.1

The low deposition energy proper of SCBD and the consequent maintaining of thenanostructure upon deposition on hard substrates have effects, for example, on the elec-tric properties of porous deposited thin films [119, 120] and exploited in different ap-plicative fields: electrodes for supercapacitors [121, 122], humidity [123] and gas sensors

Figure 3.1: Left: schematic representation of the two different (high and low-energy deposition)regimes related to clusters impact on a rigid substrate substrate [118]. Right: molecular-dynamicssimulations of thin-film growth by cluster impact at different energies [117].

27

28 3.2 Working principle of Supersonic Cluster Beam Implantation

[124] among them. Porosity of metallic nanostructured thin films also affects the wetta-bility of deposited substrates [125] and helps in improving cell adhesion [126], useful forthe fabrication of systems for diagnostics.

Recently softer substrates such as SU-8 or PMMA were used and penetration of metalnanoparticles observed. In particular electrical transport properties were analyzed andexploited for the fabrication of deformable electronic devices [114, 115]. More interest-ing are the results of Supersonic Cluster Beam Implantation of metal nanoparticles inPDMS [21] for the fabrication of highly stretchable elastomeric electrodes. The excel-lent resilience properties obatined by SCBI metal-elastomer nanocomposites representa breakthrough in the field of stretchable electronics. Moreover the biocompatibility ofPDMS is preserved by the low-energy implantation process and makes SCBI an idealtechnique for the fabrication of implantable soft electrodes for neurostimulation. Theresults on stretchable electrodes fabricated by SCBI are published in the paper HighlyDeformable Nanostructured Elastomeric Electrodes With Improving Conductivity Upon Cycli-cal Stretching reported at the end of this chapter.

3.2 Working principle of Supersonic Cluster Beam Implantation

Supersonic Cluster Beam Implantation (SCBI) is a novel technique for the implanta-tion of low-energy electrically neutral nanoparticles in soft polymers [21]. The electri-cal nature and low kinetic energy (approximately four orders of magnitude lower thanion implantation) of the nanoparticles avoid charging, carbonization and heating of theelastomer. Supersonic Cluster Beam Implantation is a high vacuum process comprisingthree main steps:

1. Metal nanoparticles are synthesized by a cluster source.

2. The flow of an inert gas (also called carrier gas), drags the nanoparticles out ofthe source and passes through a set of aerodynamical lenses aiming to collimate thecluster beam.

3. The energy gained by the nanoparticles in the previous step is sufficient for theirpenetration inside a polymeric matrix, kept on a sample holder inside a seconddifferentially pumped vacuum chamber (deposition chamber), separated from theexpansion chamber by an electroformed skimmer.

A schematic view of the SCBI apparatus is reported in Figure3.2.

3.2.1 Supersonic Expansion

The mix of inert gas and neutral metal nanoparticles produced by the cluster source isextracted by a difference of pressure between the source and the vacuum pumped in theexpansion chamber to which the source is attached. The high vacuum is necessary for theachievement of the molecular regime in which the gas and cluster beam velocity behaveaccording to the kinetic theory of gases without further approximations. In particularin a Supersonic Cluster Beam Implantation apparatus the difference between the gaspressures in the expansion chamber Pexp and in the source Psource is such to satisfy thefollowing condition:

PexpPsource

≤ 0.478 (3.1)

Supersonic Cluster Beam Implanted nanocomposites 29

Figure 3.2: Schematic view of a SCBI apparatus

under this condition the gas expansion is supersonic [116]. The term supersonic relativeto gas atoms or metal clusters in vacuum is critical and requires further clarification. Aparticles beam is said supersonic when the particles velocity in the direction of propaga-tion of the carrier gas is greater than the speed sound would have in the gas carrier itself.In a supersonic beam the particles kinetic energy is all concentrated in the propagationdirection, avoiding brownian motion of the particles due to thermal energy and thusfurther cluster aggregation. The clusters extracted from the source thus maintain theirproperties until the implantation in the polymeric substrate [127].

Nanoparticles undergoing supersonic expansion are characterized by an increase ofthe distribution of their radial velocity from the center of the beam (where it is null) tothe outer regions, and thus by a high divergence. As already seen for coating techniquesin section 2.4.2 the divergence of the metal clusters is critical for the patterning of thepolymeric substrate or the homogeneous metallization of three dimensional structures,like the grating grooves, because of the so-called shadowing effect. The divergence of acluster beam can be reduced by exploiting the different behavior of particles with differ-ent size (and so with different inertia) when subjected to an aerodynamical expansion.Small nanoparticles having a sufficiently small inertia are able to accurately follow thegas stream, while large particles trajectories are poorly affected by the gas expansion.This behavior is exploited in the aerodynamic focuser, installed between the cluster sourceand the expansion chamber and through which the mix of carrier gas and nanoparticlesexpands. Its working principle is schematically presented in Figure3.5.

The aerodynamic focuser consists in a set of steel disks, called aerodynamic lenses,through which the gas passes, undergoing subsequent expansions and compressions.The large curvature of the gas streamlines strongly affect the trajectories of the nanopar-ticles with different inertia. Small nanoparticles follow the gas stream and are expandedwith a large divergence at the exit of the focuser, large particles are not able to followthe gas and are trapped in the focuser while intermediate particles deviate from the gasstreamlines and concentrate on the focuser axis, resulting in a highly focused beam. Thediscriminant parameter quantifying the effect of particles inertia is the Stokes number,


Figure 3.3: Detailed scheme of the effect of nanoparticles inertia when dragged through a smallaperture by the carrier gas expansion.

defined as [128, 129]:St = τ

v

d(3.2)

where v is the clusters mean speed, d the diameter of the drill in the aerodynamic lensand τ the relaxation time, representing the characteristic time the particle takes to changeits trajectory and calculated as:

τ =d2pρpCc

18η(3.3)

where dp is the diameter of the nanoparticle, ρp its density, η the viscosity of the gas. theCunningham factor Cc corrects the formula for sub-micrometric particles and dependson the free mean path of the gas atoms and on the particle diameter. The cluster beamexiting the aerodynamic focuser is characterized by unitary Stokes number particles atthe center and by particles with decreasing Stokes number towards the borders. So theaerodynamic focuser acts also as a particles size selector, filtering only the particles witha given size and speed. The high collimation (divergence lower than 1 degree) of thecluster beam exiting the aerodynamic focuser allows the use of stencil masks, not in directcontact with the polymer surface, for the micropatterning of the elastomeric substratewith a high lateral resolution [116, 22], avoiding the use of chemicals needed for thedevelopment of lift-off masks and the shadowing effect typical of traditional coatingtechniques like PVD.

3.2.2 Implantation in soft polymers

Clusters accelerated in a supersonic regime by the carrier gas expansion gain an energyof approximately 0.5-2.0eV/atom, sufficient for their implantation inside the polymericmatrix but not in hard substrates like glass or silicon. The local decrease of the poly-mer density upon implantation of the first nanoparticles suggest an easier implantationof further nanoparticles at higher penetration depths thanks to the clearing the way ef-fect already observed for ion implantation in hard substrates [130, 131, 132] and can beconsidered one of the key ingredients of the implantation mechanism in SCBI.

The low kinetic energy gained by the nanoparticles during the supersonic expansionguarantees their implantation in elastomers while maintaining the chemical and physi-cal properties of the elastomeric substrate unaltered. Considering the number of atomsNat reaching the polymer surface per second, the surface power density Psurf for clusterimplantation can be calculated as Psurf = Nat ·Ek, where Ek is the kinetic energy of the


atoms. The power density is of the order of some µW · cm−2 in the case of SCBI, not suf-ficient for a detectable heating of the elastomeric substrate above RT. As a comparison,the power density for ion implantation is much larger than SCBI, of the order of tenthsof W [133].

3.2.3 Equivalent thickness and filler volume fraction

The determination of the amount of nanoparticles implanted in the polymer by SCBI isnot trivial. In PVD or ion implantation directly measurable parameters like the depo-sition thickness or the implanted atoms dose are respectively used. In SCBI howevermetal is implanted in the polymer, thus the concept of deposited thickness has no sense.The measure of the equivalent thickness represents an effective way for the determinationof the amount of implanted nanoparticles in the polymeric matrix. A half-masked hardsubstrate (silicon or glass) is placed next to the polymeric substrate to implant, inter-cepting the same amount of nanoparticles implanted in the elastomer. Nanoparticlesreaching the hard substrate are deposited on its surface instead of getting implanted,leading to a two-dimensional growth of a nanostructured layer on the non-masked freesubstrate surface. The difference of height between the deposited and the masked hardsubstrate regions, measured with a profilometer or an AFM, defines the equivalent thick-ness teq . Equivalent thickness can be also monitored in situ by a quartz microbalance,useful for the live control of the implantation process but whose measured value doesnot directly coincide with the equivalent thickness measured after the implantation pro-cess on the rigid substrate, because of the porosity typical of nanostructured depositedfilms, as further discussed in section 3.1.

From the measure of the equivalent thickness and the penetration depth obtainableby TEM images the filler volume fraction f of the nanocomposite can be obtained, definedas the ratio between the total volume of the metal clusters implanted (the filler) Vclusterand the total volume of the nanocomposite Vnc:

f =VclusterVnc

(3.4)

The non-trivial distribution of the nanoparticles inside the nanocomposite makes thefiller volume fraction difficult to measure. If a uniform nanoparticles distribution in-side the nanocomposite is assumed, the filler volume fraction can be rewritten, in firstapproximation, as follows:

f =teqdnc

(3.5)

where dnc represents the maximum penetration depth of the nanoparticles in the poly-mer matrix.

The filler volume fraction is important in determining the macroscopic physical prop-erties of the whole nanocomposite. Considering the electrical properties,the knowl-edge of the volume fraction is important in order to model the percolation curves andso to extract physical parameters describing the electrical behavior of the metal-polymernanocomposite [21, 118]. The volume fraction also represents a critical parameter for thedescription of the light absorption response of the nanocomposite, as it will be shown inpart III of the present work.


Figure 3.4: Vertical trajectories (top graphs) and velocity (bottom graphs) of the nanoparticles withdifferent kinetic energies in the entangled melt (left) and cross-linked (right) polymeric substrates(from [134]).

3.2.4 Molecular dynamics simulations of SCBI in PDMS

The implantation mechanism of electrically neutral metal nanoparticles by SCBI in Poly-dimethylsiloxane (PDMS) was theoretically and extensively studied by means of molec-ular dynamics simulations [134]. The PDMS substrate was simulated with a force fieldaccounting for all the possible vibrational and rotational modes of the PDMS moleculessubjected to a 9-6 Lennard-Jones potential and in two different configurations: entangledmelt and cross-linked. Entangled melt polymer consists in 100 monomers chains recip-rocally twisted to form the polymeric matrix, while the cross-linked polymer consistsin polymeric chains covalently bonded by means of cross-linking molecules (Tetrakis(dimethylsiloxy) silane in the case of PDMS). The simulated PDMS substrates were suffi-ciently large for an efficient dissipation of the pressure waves generated upon the clusterimpact and stabilization of the cluster position inside the substrate. The substrate wasfixed at the bottom and thermalized with a Nose-Hoover thermostat at 300K.

Metal clusters, with a diameter of 3nm, are shot toward the polymer at energies of0.5eV/atom, 1eV/atom and 2eV/atom, in order to study the effect of the cluster energyon the implantation process. The trajectories and velocities of the particle along thevertical direction were recorded and plotted in Figure3.4.

Four phases in the cluster implantation process, named S1, S2, S3 and S4, can be welldistinguished. The response of the polymer after the implantation, both in the entangledmelt and the crosslinked substrate, dominate the behavior of the metal particle after itsimplantation. During step S1 the cluster moves freely in vacuum with a linear uniformvelocity, depending on the implantation energy, before reaching the polymer surface.During step S2 the cluster impacts on the polymeric surface and gets implanted, loos-ing its kinetic energy because of the friction forces between the nanoparticle and the


PDMS matrix (the penetration depth reaches its maximum and the velocity decreases tozero). In step S3 the response of the PDMS takes place: the PDMS is compressed uponthe cluster impact, the pressure wave reaches the fixed polymer atoms at the bottomof the simulation cell and is reflected back to the surface. The nanoparticle undergoingthe reflected pressure wave is pushed upward toward the surface and the penetrationdepth slightly decreases. A stable configuration in which the penetration depth remainsunchanged and the cluster velocity reaches a null value is achieved in step S4.

The effect of the implantation energy is clear in Figure3.4: by increasing the clusterenergy the penetration depth increases as well and the S2 and S3 steps are longer, as ex-pected. More interesting the differences between the entangled melt and the crosslinkedsubstrates. In the latter case the penetration depth is smaller and the polymer responseafter implantation stronger. This is compatible with the breakage of bonds between thepolymeric chains in the case of the crosslinked substrate or their untwisting in the entan-gled melt polymer. The bonds in crosslinked polymers are mainly covalent and 8% morethan the entangled melt configuration, in which the polymeric chains are dispersed andthe bonds are due to much weaker electrostatic forces. The penetration depth vs. implan-tation energy follows a linear trend both in the case of entangled melt and the crosslinkedconfigurations, with a slope of 7nm/eV and 6nm/eV respectively. Moreover a swellingof the polymer surface is observed in both cases and for all the implantation energies. Inthe case of the crosslinked matrix the swelling is approximately 3nm independently ofthe implantation energy, larger than in the case of the entangled melt polymer, becauseof the larger amount of broken covalent bonds and to a stronger elastic response of thesubstrate.

The formation of pressure waves and the breakage of bonds between the polymerchains are not the only effects due to cluster implantation in soft polymers. A local heat-ing of the polymer around the implanted particles and of the particle itself is observed.Polymer atoms next to the particle are subjected to an increase of temperature from 300Kup to 600K but the mean temperature increase in the the first superficial 50nm thick slabof the polymer is approximately 30-50K, not affecting its global physical properties.

Cluster implantation in soft polymers also creates a crater, slowly recovered in timedue to chains rearrangement. Surface morphology of the implanted polymer is affectedby the formation of these craters. Experimental measures of the surface roughness werecompared with the depth of the craters and the roughness measured by simulated AFMexperiments, as explained in [22].

In molecular dynamics simulations only a single particle implantation is considered.A multiparticles implantation simulation requires a much heavier computing effort (forthe simulations analyzed 65536 CPU hours were required for each impact). Howeverthe implantation of the first nanoparticles facilitates the implantation of further particlesthanks to the clearing the way effect previously discussed.

3.3 Supersonic Cluster Beam Implantation setup

In section 3.2 a general description of the working principle and main advantages ofthe SCBI respect to other nanocomposite synthesis techniques were given, without anydetail on the cluster source, since SCBI is compatible with different available sources,and of the SCBI apparatus.

The SCBI apparatus used in this thesis work is equipped with a Pulsed MicroplasmaCluster Source (PMCS) [135, 116] for the synthesis of metal nanoparticles. Metal atomsare sputtered by a metal rod through ionized inert gas atoms, and aggregate inside thesame gas forming electrically neutral metal nanoparticles. A more detailed description

34 3.3 Supersonic Cluster Beam Implantation setup

Figure 3.5: (a) Exploded view of the aerodynamical focuser used in the SCBI apparatus. (b)Scheme of the streamlines of nanoparticles with St ' 1 through subsequent focuser stages.

of PMCS will be given in the following. The mix of gas and nanoparticles then expandsin the expansion chamber, to which the PMCS is connected, through the aerodynamicalfocuser. The high vacuum kept in the expansion chamber (approximately 10−6torr) by aturbomolecular pump having a flow rate of 1900l · s−1, is responsible for the extractionand supersonic acceleration of metal nanoparticles toward the elastomeric substrate. Theaerodynamical focuser, schematically reported in Figure3.5 is made of a series of cylin-ders (the focuser stages) separated by aerodynamical lenses. These lenses consist in steeldisks with a central 2mm diameter drill (1mm diameter for the last lens). The abruptchanges of direction of the gas-nanoparticles mixture expanding through each aerody-namical lens allow both a selection of the nanoparticles with a certain size (in particularnanoparticles characterized by St < 1, as extensively discussed in section 3.2.1) and theircollimation on the axis of the focuser.

The central part of the collimated cluster beam is intercepted by a 2mm small aper-ture for the further spatial filtering of the nanoparticles. This device, called skimmer ischaracterized by a low conductance and a swallow tail shape in order to keep a differ-ential vacuum between the expansion and the deposition chambers, and limits the su-personic shock waves caused by the supersonic expansion respectively. Supersonicallyaccelerated gas atoms exiting the aerodynamical focuser expand with a high divergence,dragging metal nanoparticles with St� 1. As a consequence only the central part of thebeam, consisting of metal nanoparticles with St ' 1 is able to pass through the skimmerand to enter the deposition chamber.

Deposition chamber is kept at high vacuum (approximately 10−5torr by a secondturbomolecular pump with a flow rate of 500l · s−1. Here nanoparticles get implanted inthe elastomer, kept on a sample holder connected to a computer-controlled manipulatorand able to move orthogonally to the cluster beam direction for the implantation onlarge areas. The sample holder is equipped with a quartz microbalance for the real-time monitoring of the implantation rate and, if needed, with a heater for the controlledheating of the sample during the implantation process.

The section of the nanoparticles beam intercepted by the sample holder in the de-position chamber consists in a circle with a diameter of few centimeters. This circularspot is inhomogeneous, since the nanoparticles density decreases from the center to the


Figure 3.6: Front and rear schematic representation of the sectioned PMCS.

borders. The movable sample holder on which the elastomeric substrate is mountedguarantees a homogenous implantation on larger areas, up to 20x20cm2. The relativepath followed by the nanoparticles beam through the movement of the substrate is re-peated many times until the achievement of the wanted equivalent thickness throughoutthe entire sample.

3.3.1 Pulsed Microplasma Cluster Source (PMCS)

The source (depicted in Figure3.6) consists in a ceramic body with a cylindrical cav-ity and two openings at the opposite sides of the cavity. At the entrance a solenoidalpulsed valve allows the injection of an inert gas, expanding and concentrating around arod composed of the wanted nanoparticles material (a conductive metal or metal alloy)inserted in the source through a third aperture. Immediately after the valve, a drilledcopper disk acts as counter electrode in the functioning mechanism of the source. Atthe opposite opening of the ceramic body, a nozzle characterized by a low conductanceis needed in order to maintain a differential pressure between the source and the vac-uum chamber to which the source is applied. The whole system is inserted in an alu-minum support electrically connected to the copper disk and grounded. A motorizedmechanism is connected to the aluminum support for the rotation of the rod during theoperation of the source.

Like most of the cluster sources, the working principle of the PMCS can be dividedin four phases, as shown in Figure3.7: gas injection, vaporization of the metal target,nanoparticles condensation and nanoparticles extraction. The pulsed valve is openedfor few hundreds microseconds, letting the inert gas (typically Argon at a pressure of40-50 bar) entering the ceramic cavity and concentrating around the metal rod, becauseof aerodynamical effects due to the shape of the cavity and the rod diameter. After 80-100µs from the valve opening, an electrical discharge of about 800V between the rod(acting as a cathode) and the copper disk (the anode) with a duration of some hundredsmicroseconds takes place. The discharge ionizes the gas atoms that are thus attractedby the biased metal rod and gain sufficient energy to sputter metal atoms from the roditself. Sputtered metal atoms thermalize and condensate in the gas forming electrically

36 3.3 Supersonic Cluster Beam Implantation setup

Figure 3.7: PMCS working principle. The inert carrier gas is injected in the source ceramic cavity(a) and concentrates around the metal rod. A high voltage is applied between the rod and theanode, and an electrical discharge sputters metal atoms from the rod (b). The metal atoms ther-malize and aggregate in the gas forming electrically neutral metal nanoparticles (c) and the mix ofgas and nanoparticles is extracted by a difference of pressure with the vacuum chamber to whichthe source is applied (d).

neutral metal nanoparticles, with sizes ranging from few up to tens nanometers, dispersedinside the high pressure inert gas. The source is applied to a high vacuum chamber andthe difference of pressure, kept by the low conductive nozzle at the exit opening of thesource, causes the expansion of the gas dragging the synthesized nanoparticles. Thewhole process is repeated at a frequency varying between 1 and 10Hz, producing highintensity nanoparticles pulses.

The working parameters of the source are extremely important and must be adjustedin order to control the nanoparticles sizes and extraction energy according to the rodmaterial. During the source operations the rod is kept in rotation in order to guaranteea homogeneous erosion and a regular and constant functioning of the source, critical forsynthesis of nanocomposites with optimal and repeatable properties.

Similar cluster sources are available, differentiating only in the metal vaporizationmechanism. In Pulsed Laser Vaporization Sources (PLVS) [136, 137] a laser beam im-pinging on the metal target erodes metal atoms in a limited area, ensuring the time sta-bility of the process. Intensity of the cluster beam is the major problem. In Pulsed ArcCluster Ion Sources (PACIS) [138] metal atoms are extracted from the target by meansof an arc discharge, allowing a larger intensity. The strong modification of the targetgeometry and gas dynamics affects the time stability (see [135] and references therein).In the PMCS the gas concentrated around the metal rod allows localizing the dischargein a well defined region of the target, aiming to achieve both intensity typical of electricarc based sources and the time stability of laser based sources.


3.4.1 Size distribution of metal nanoparticles

The size distribution of the nanoparticles passing through the skimmer and implanted inthe sample was investigated by TEM. Metal nanoparticles (in particular silver nanopar-ticles used for the metallization of elastomeric optical devices in this work) were de-posited on a Formvar-coated nickel TEM grid metallized with a Carbon submonolayerfor three seconds in order to obtain isolated and uniformly dispersed nanoparticles onthe grid. During the analysis of the images only particles with an eccentricity greaterthan 0.7 were considered, in order to exclude cluster aggregates. The results for silver,presented in Figure3.8, show a median nanoparticle size of 9.05 ± 0.53nm. A similarcharacterization was carried for gold nanoparticles in [22]. In both cases nanoparticlessize distribution is small enough to avoid particles scattering phenomena.

Figure 3.8: From left to right: TEM image of Ag nanoparticles deposited on the Formvar layer,Probability distribution function (PDF) of the nanoparticles count and the corresponding Cumu-lative Distribution Function.

3.4 Morphological and mechanical characterization of SCBI Ag/PDMSnanocomposites

Morphological and mechanical properties of metal-polymer nanocomposites synthe-sized by SCBI are analyzed in light of the requirements needed for elastomeric opticaldevices, mainly in terms of light scattering, surface roughness and stretchability (or elas-ticity).

3.4.2 Penetration depth of metal nanoparticles

The penetration depth of metal nanoparticles in PDMS can be investigated by TEM.PDMS substrates are implanted for few seconds by SCBI, then cut by Cryoultramicro-tomy in approximately 100nm thick slices and laid on a Formvar-coated TEM grid. Silvernanoparticles, used in the following of this work for elastomeric optical devices, are wellvisible in the TEM images (like, for example the one shown in Figure3.9) and are im-planted up to a penetration depth of approximately 100nm. A similar characterizationfor gold nanoparticles was presented in [22]: here a penetration depth of approximately180nm was observed, independently on the implanted equivalent thickness. The attain-able penetration depth is thus similar for different metal nanoparticles and of the orderof tens of nanometer.

38 3.4 Morphological and mechanical characterization of SCBI Ag/PDMS nanocomposites

Figure 3.9: TEM image of the Ag/PDMS nanocomposite

3.4.3 Elasticity of SCBI nanocomposites

Preliminary results on the surface Young modulus of the SCBI nanocomposite were ob-tained by AFM nanoindentation experiments on a nanocomposites library character-ized by different equivalent thicknesses (and thus volume fraction) and reported in Fig-ure3.10. The contact region of AFM indentation curves on the nanocomposite surface,acquired with proper spherical micrometric colloidal probes, were analyzed exploitingthe Hertz contact model describing the interaction mechanism between a rigid sphereand an elastic surface. The measured Young modulus varies from 2.5MPa of the barePDMS up to approximately 12MPa for a gold implanted equivalent thickness of 100nm(corresponding to a volume fraction of about 0.20 if the polymer swelling upon implan-tation is considered), two orders of magnitude lower than in the case of ion implanta-tion. These nanoindentation measurements involve the first hundreds nanometers of theimplanted polymer, not considering the remaining part of the PDMS substrate that ex-tends for tens of micrometers. Very preliminary results of the Young modulus obtainedby stress-strain simulated experiments on nanocomposite samples (nanocomposite plussupporting PDMS) prepared by molecular dynamics are comparable to the results ob-tained by AFM mesurements.

The much lower increase of Young modulus achieved by SCBI respect to other higherenergy implantation techniques can be explained again with the implantation mecha-nism occurring in the two cases. SCBI only causes a slight local increase of the tempera-ture of the polymer, leading to debonding of the polymeric network. Unlike ion implan-tation, low energies involved in the SCBI process do not alter the structure and chemi-cal composition of the polymeric chains and thus their elasticity, as predicted by [113].Therefore the increase of the stiffness observed for SCBI could be explained just by con-sidering the effect of the filler in the polymer, as predicted by Guth equation [88, 89]. Asa proof of this hypothesis, preliminary experimental data follow the same trend reportedin Figure2.9 and 3.10 for different aspect ratios, as expected because of the formation ofaggregates.


Figure 3.10: Comparison between AFM data and expected data from Guth equation [88, 89] fordifferent filler aspect ratios α.

3.4.4 Surface roughness

Surface morphology is critical for the exploitation of metal-polymer nanocomposites foroptical applications. Surface roughness is responsible for the scattering of the imping-ing light and must be therefore minimized and controlled. Moreover three dimensionalsub-micrometric features like the grooves of a diffraction grating must be perfectly re-produced by the metallization process. The effect of SCBI of metal nanoparticles on themorphology of PDMS in which they are implanted was extensively studied in the paper[22] titled Patterning of gold-polydimethylsiloxane (Au-PDMS) nanocomposites by supersoniccluster beam implantation, published on Journal of Physics D: Applied Physics, entirely re-ported at the end of the chapter. In this paper PDMS was implanted with gold nanopar-ticles, however few differences are observed by implanting different metal nanoparti-cles. Morphological measurements on Ag/PDMS nanocomposites are presented as acomparison

Au/PDMS nanocomposites surface roughness characterized by different equivalentthicknesses was investigated by AFM investigation. The results show an increase ofthe surface roughness, up to approximately 10nm for an equivalent thickness of 100nm.From an optical point of view this translates to a very low nanocomposite light scatter-ing. Recalling the results of equation 2.3, reported in Figure2.3, a roughness of 10nm isresponsible for a constant light scattering less than 6%, slowly decreasing with the lightincident angle.

Moreover from the measure of the step between the bare PDMS and the nanocom-posite, a swelling of the implanted polymer is observed, as reported by the moleculardynamics simulations. However the swelling phenomenon does not affect the lateralresolution of a patterned region, the gradient remaining less than 1µm in extension.Swelling is supposed to affect the geometry of implanted three-dimensional structureslike, for example, the diffractive structure of a grating, but since the sample is homoge-neously implanted, the profile remains unaltered, as it will be shown in the next chapter.

These results regard Au nanoparticles with a mean size of 3.9 ± 1.4nm implanted in

40 3.4 Morphological and mechanical characterization of SCBI Ag/PDMS nanocomposites

Figure 3.11: 2000µm x 1000µm AFM topographic maps of Ag/PDMS (a) and Au/PDMS (b)nanocomposites with comparable equivalent thicknesses (87nm and 96nm respectively).

PDMS. However optical devices considered in the following chapters of this work, arefabricated by implanting Ag nanoparticles, with a larger mean size, in the same poly-mer. Nevertheless the attainable RMS roughness is similar or better than Au/PDMSnanocomposites. The surface roughness of a Ag/PDMS nanocomposite implanted withan equivalent thickness of 87nm was measured by AFM and results in 6.6±0.3nm, muchlower than the surface roughness of a Au/PDMS nanocomposite implanted with a sim-ilar equivalent thickness (teq = 96nm, roughness 10.1 ± 0.4nm in the case of gold). Thetopographic maps of the two samples are reported in Figure3.11. According to these re-sults the fraction of light scattered by the Ag/PDMS nanocomposite is even lower (fewpercents) than in the case of Au/PDMS.

© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim4504

www.advmat.dewww.MaterialsViews.com

wileyonlinelibrary.com

CO

MM

UN

ICATI

ON

Adv. Mater. 2011, 23, 4504–4508

G. Corbelli , C. Ghisleri , M. Marelli , Prof. P. Milani CIMAINA and Physics Department Università degli Studi di Milano Via Celoria 16, I-20133 Milano, Italy E-mail: [email protected] Dr. L. Ravagnan WISE s.r. l. Via Boschetti 1, I-20121 Milano, Italy E-mail: [email protected]

DOI: 10.1002/adma.201102463

Many applications in biomedicine, prosthetics, wearable elec-tronics, and robotics require the integration of electronic, optical, and actuation capabilities on soft and conformable polymeric substrates. [ 1 ] Much progress has been made in this area, in particular in the fabrication of circuits and devices on fl exible substrates [ 2 ] and those that utilize mass-production manufacturing processes to produce fl exible solar cells, [ 3 ] fl ex-ible displays, [ 4 ] smart clothing, [ 5 ] sensors, and actuators. [ 6 , 7 ] For example, Rogers and co-workers recently reported a major achievement towards stretchable and high-performance micro-electronics capable of tolerating very large levels of strain by integrating ultrathin silicon-based electronic circuits into elas-tomeric substrates. [ 8 ] This platform was used for the fabrication of smart surgical tools that featured enhanced performances. [ 9 ]

Despite these achievements, stretchable electrodes consisting of metallic paths on elastomeric susbtrates are still plagued by drawbacks and failures that prevent their use, especially for biomedical applications. Implantable devices for neurostimu-lation and neuroprosthetics [ 10 ] can strongly enhance their per-formances and enlarge their fi eld of application by the possi-bility of printing metallic microcircuits on biocompatible and conformable substrates. Conceivably, the treatment of patholo-gies including Parkinson’s disease, essential tremor, dystonia, chronic pain, treatment-resistant depression, cluster head-aches, and epilepsy [ 11 ] could benefi t signifi cantly from the use of stretchable electrodes.

Efforts to fabricate stretchable metallic circuits and electrodes are concentrated on the direct metallization of polydimethylsi-loxane (PDMS), which couples biocompatibility with mechanical properties and machinability suitable for rapid prototyping. [ 12 ] At present the metallization of PDMS to produce micrometric and well-defi ned conductive pathways is obtained by metal vapor deposition (e.g., Au, Ag, Pt, Pd) [ 13 ] or metal ion implan-tation. [ 14 , 15 ] These techniques are in principle quite straightfor-ward, however, in practice, the use of conformable electrodes produced by the former approach is hampered by delamination of the conducting layers even at very low deformations. The use of adhesion layers such as Cr or Ti or the treatment of the

PDMS surface with oxygen plasma improve the performance, however the resilience is not adequate for the majority of appli-cations. [ 13 , 16 ] Several attempts have been made to overcome the deterioration of evaporated layers by the deposition of tortuous wave-shaped traces [ 17 ] or by the use of prestretched PDMS sub-strates, [ 18 ] but these solutions have considerable limitations in terms of maximum applicable strain, the direction of the toler-able strain, the density of attainable circuits, and scalability.

Better adhesion can be obtained by metal ion implantation where noble metal ions are implanted at 4–10 keV atom − 1 into a polymeric substrate [ 14 , 15 ] forming a metal–polymer nanocom-posite conductive layer below the substrate surface. [ 14 ] Stretch-able electrodes fabricated by this technique show adhesion and electrical conductivity degradation rates upon cyclical stretching that are substantially superior to evaporated electrodes but, nevertheless, this technique has several drawbacks, since it induces charging and carbonization of the insulating polymeric substrate.

Here we demonstrate that stretchable and compliant elec-trodes on PDMS can be effi ciently fabricated by the implanta-tion in the elastomeric substrate of neutral metallic clusters aerodynamically accelerated by a supersonic expansion. Super-sonic cluster beam implantation (SCBI) is an effective method for the microfabrication of conductive circuits on thermoplastic polymers, [ 19 ] it is applicable at room temperature (without any heating of the samples), and it does not induce any charging or carbonization of the polymeric substrate. Using SCBI, we fabricated deformable elastomeric electrodes that can withstand deformation of 40% for more than 10 5 cycles and that decrease their resistance upon cyclical stretching.

SCBI consists in directing a highly collimated beam of neu-tral (i.e., with zero electric charge) metallic clusters ( Figure 1 a) with a size distribution range of 3 nm to 10 nm and kinetic energy of about 0.5 eV atom − 1 towards a polymeric substrate. Although the kinetic energy per atom of the clusters is four orders of magnitude lower than in the case of ion implanta-tion, clusters (made of several thousand atoms) have suffi cient inertia to penetrate into the polymeric target, which is kept at room temperature (RT), and to form a nanocomposite layer, while avoiding charging and carbonization of the polymeric substrate. [ 19 ]

Figure 1 b,c show transmission electron microscopy (TEM) images of cross sections of a PDMS substrate implanted at RT with Au clusters at corresponding equivalent thicknesses [ 19 ] of 35 nm and 210 nm, respectively. Remarkably, the penetration depth of the Au clusters in PDMS (90–136 nm, Figure 1 b,c) is approximately twice the penetration depth that was observed for a Pd cluster in poly(methyl methacrylate) (PMMA) at RT (approximately 50 nm [ 19 ] ). This difference is expected because of the lower hardness of PDMS in comparison with PMMA

Gabriele Corbelli , Cristian Ghisleri , Mattia Marelli , Paolo Milani , * and Luca Ravagnan *

Highly Deformable Nanostructured Elastomeric Electrodes With Improving Conductivity Upon Cyclical Stretching

4505


© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com

CO

MM

UN

ICATIO

N

Adv. Mater. 2011, 23, 4504–4508

percolation threshold s c , which is defi ned here, operatively, as the equivalent thickness at which the sample conductance exceeds 10 − 10 Ω − 1 . [ 19 ] On glass, s c is about 5 nm, corresponding to a coverage of approximately 50% of the substrate by the clusters, which have a mean size of about 8 nm, and thus con-sistent with a growth of a cluster-assembled fi lm on the surface of the substrate. [ 20 ] In PMMA the percolation is reached for s c of about 10 nm, clearly indicating the penetration of the cluster in the polymer matrix, consistent with previous results. [ 19 ]

and is furthermore supported by the characterization of the evolution of the electrical transport properties, i.e., the con-ductance, of the samples as a function of the amount of depos-ited clusters, i.e., the equivalent thickness. Figure 1 d shows the percolating behaviors measured by depositing clusters on three different substrates simultaneously: glass (in which clus-ters are not implanted), PMMA, and PDMS. As can be clearly seen, on all three substrates the characteristic evolution from insulating to conductive behavior occurs beyond a critical

Figure 1 . a) Schematic view of the apparatus used for SCBI (not to scale). The collimated supersonic beam of neutrally charged Au nanoparticles (produced by the cluster source, mean cluster size 8 nm) impact the PDMS substrate forming an Au/PDMS nanocomposite layer. b,c) TEM images of cross-sections (cut by cryo-ultramicrotomy, thickness 300 nm) of the produced Au/PDMS nanocomposite samples obtained at RT by implanting an equivalent thicknesses of clusters of 35 nm and 210 nm, respectively. d) Measurement of the evolution of the conductance on three substrates (glass, PMMA, and PDMS) simultaneously exposed to the cluster beam. Measurements are shown as a function of the equivalent thickness, which was measured by a quartz microbalance. [ 19 ]

4506


© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimwileyonlinelibrary.com

CO

MM

UN

ICATI

ON

Adv. Mater. 2011, 23, 4504–4508

(i.e., on glass), which is consistent with the nanocomposite nature of the material.

We tested the performance of a conductive PDMS/gold nano-composite obtained by SCBI against extensive uniaxial strain cycles. To do this we produced the nanocomposite fi lm across the gap between two evaporated electrodes on a PDMS membrane ( Figure 2 a, see details in the Experimental Section). The sample was then mounted on a custom-built, computer-controlled, motorized uniaxial stretcher, allowing automatic acquisition of the nanocomposite electrical resistance ( R ) with cyclic strain ( ε ). At each strain cycle, the maximum applied strain was 40%. The electrical resistance was recorded during “measuring” strain cycles at an elongation rate of 20 μ m s − 1 , alternating with a series of “fatigue” strain cycles, at a faster elongation rate (200 μ m s − 1 ).

Figure 2 b shows the resulting stretching cycles recorded during cycle 2, 10, 100, 1000, 10 000, and 50 000 on a representa-tive sample. The same evolution was observed over multiple experiments on different samples.

Remarkably, on PDMS the fi lm remains insulating up to a s c value of 75 nm. Qualitatively, a value of s c that is higher for PDMS compared with PMMA confi rms the larger penetra-tion of the gold clusters observed by using TEM. Nevertheless, quantitatively, the s c value of 75 nm for PDMS would sug-gest a penetration depth much higher than the observed one. This discrepancy can be understood in light of the fact that at RT gold clusters have a low but not zero mobility in PDMS, which has a glass transition temperature of about –120 ° C, [ 21 ] and this allows them to aggregate due to an Ostwald ripening process. [ 22 ] The same does not happen for PMMA because it has a glass transition temperature above RT. The substantial shift in the percolation threshold can therefore be ascribed to the formation, in the fi rst stages of the nanocomposite growth, of dendritic aggregates of gold clusters in the PDMS matrix. Finally, it is worth noting that the conductance of the PDMS/gold nanocomposite is always lower, by at least one order of magnitude, than the conductance of the surface-deposited fi lm

Figure 2 . a) Schematic view of the experimental setup used for the characterization of the evolution of the transport properties during uniaxial strain cycles. b) Electrical resistance as a function of applied strain recorded on the Au/PDMS nanocomposite fi lm (size of the suspended portion of the sample subjected to strain: 30 mm wide by 5 mm long) produced by SCBI, during cycle 2, 10, 100, 1000, 10 000, and 50 000 (maximum elongation: 40%). During measuring cycles the samples were stretched at an elongation rate of 20 μ m s − 1 . c) Electrical resistance at 0% strain ( R in ) and at 40% strain ( R fi n ) as a function of the number of stretching cycle to 40% strain. d) Electrical resistance as a function of applied uniaxial strain. The failure at 97% strain is due to the PDMS substrate mechanically breaking.

4507


© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim wileyonlinelibrary.com

CO

MM

UN

ICATIO

N

Adv. Mater. 2011, 23, 4504–4508

stretchable substrates. The approach is based on the accelera-tion of metallic clusters by a supersonic expansion and their implantation in PDMS substrates to form a conductive nano-composite. The produced structures are characterized by their superior capability to sustain very large deformation with electrical conductance, which improves with cyclical deforma-tion. This approach overcome the many limitations typical of standard metallization approaches, in terms of performances (delamination, etc.) and processing (sample heating, electrical charging, carbonization, use of solvents, and use of adhe-sion layers). Microfabrication of conductive nanocomposite patterns on elastomers provide new perspectives for stretch-able and conformable electrodes for biomedicine and smart prosthetics.

Experimental Section The cluster beam used for the implantation was produced using a pulsed microplasma cluster source (PMCS) [ 25 ] equipped with a cavity in which a solid target made of the metals desired for the clusters (Au in the present case) was vaporized by a localized discharge supported by a pulsed injection of an inert gas (He or Ar) at high pressure (Figure 1 a). The vaporized metal atoms aggregated in the cavity to form metal clusters, which were then driven out of the source cavity by the expansion of the inert gas through a nozzle to a vacuum chamber (expansion chamber, at a pressure of about 10 − 5 bar). Because of the high pressure difference between the source cavity and expansion chamber and the exploitation of aerodynamic focusing effects, the neutrally charged clusters were emitted from the source in the form of a highly collimated beam with a divergence lower than 1 ° and with the kinetic energy and mass distribution previously described. [ 25 ] The central part of the cluster beam was then allowed to enter a second vacuum chamber (deposition chamber) through a skimmer and it was intercepted by the polymeric substrate, which was supported by a motorized substrate holder. The holder allowed movement of the substrate in two directions orthogonal to the cluster beam axis (rastering), permitting the cluster implantation on an arbitrarily wide polymeric substrate. The substrates were exposed for ≈ 180 min at a deposition rate of 0.2 Å s − 1 .

The Au targets (99.99% pure) were purchased from the 8853 S. p. a. company, and the PDMS substrates were prepared by spin-coating an ≈ 80 μ m thick layer of the polymer precursor (Sylgard 184 Silicone Elastomer Kit from Dow Corning, mixed in a 10:1 ratio) on a glass slide covered by a thin layer of PMMA to reduce the adhesion of the elastomer fi lm, at 1000 rpm for 60 s. After spin-coating, the polymer fi lm was cured by heating at 100 ° C for 35 min, as indicated by the producer.

During cycle 2 the nanocomposite electrical resistance grew almost linearly from an initial value of 23 Ω (at 0% strain, R in ) up to 420 Ω when the maximum strain was reached, i.e., at 40% strain, R fi n , recovering its initial value when the strain was released. Remarkably, as the number of strain cycles grew, the increase of R with ε remained monotonous also after 50 000 cycles, maintaining an almost triangular response. This is the fi rst signifi cant departure from the typical behavior observed for evaporated metal fi lms on polymers, where the formation of cracks in the rigid superfi cial fi lm induces a non-monotonous increase in R with ε after a few thousand strain cycles with max-imum applied strain much lower than 40%. [ 13 ] Furthermore, the value of R in for the nanocomposite exhibited only a slow increase as the number of cycle increased and, most remarkably, the value of R fi n progressively decreased. As shown in Figure 2 c, the value of R fi n became almost half the initial value after 50 000 strain cycles, which is completely at odds with the usual huge increase (by orders of magnitudes) of R fi n observed for evapo-rated metal fi lms. [ 13 ] Those differences can be explained by the nanocomposite nature of the conductive fi lm, where the increase of the R with strain application is mainly due to the increase in the mean distance between the metal nanoparticles, which causes the opening of several percolating paths. [ 14 ] Nevertheless, when strain is released the percolating paths are reformed and the electric conductivity is recovered. Moreover, the repetition of uniaxial strain cycles allows the nanoparticles embedded in the nanocomposite, which have a nonzero mobility in the polymer, to progressively reorganize themselves and aggregate, leading to a percolating network that is less affected by the strain and consequently to the decrease of R fi n . Furthermore, the nano-composite is not characterized by the supervening of an abrupt electrical failure at a critical applied strain as is the case for evaporated metal fi lms, [ 23 ] but to a progressive increase in R . This is shown in Figure 2 d, where a “break test” is performed by subjecting the fi lm to growing strain up to the electrical or mechanical failure of the sample, i.e., the breaking of the polymer membrane. For the tested fi lm, the occurrence of an electrical failure before mechanical failure was never observed and the fi lm could withstand up to 97% strain.

SCBI is not only an effi cient method for the production of stretchable and durable metallic circuits but it also functions as a micropatterning tool. By exploiting the high collimation typical for supersonic cluster beams, [ 24 ] micrometric patterns can be easily obtained by interposing a stencil mask in front of the PDMS substrate ( Figure 3 a). Figure 3 b shows an example of a gold micropattern obtained using a TEM grid that is not in contact with the substrate as a stencil mask. Figure 3 c shows the same micropattern under stretching to 35% strain. No delamination is observed, although a few superfi cial micro-cracks are present. Scanning electron microscopy (SEM) images showing in detail the cracks formed after the uniaxial stretching of the sample are presented in the Supporting Infor-mation and show the superfi cial nature of the cracks, which infl uences the electrical conductivity of our material in a mar-ginal way. The fabrication process consists of one step and it does not include any wet chemical treatment, thus avoiding surface contamination.

In summary we have demonstrated a novel fabrication process to obtain metallic electrodes and micropatterns on

Figure 3 . a) Schematic view of the process of stencil mask lithography: a TEM grid from Agar (G2760N) is interposed between the cluster beam and the PDMS substrate. b) A microscopy image of the obtained Au/PDMS nanocomposite micropatterns (the bright hexagons). c) A micro-scopy image of the same micropatterns stretched to 35% strain.

4508


© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheimwileyonlinelibrary.com

CO

MM

UN

ICATI

ON

Adv. Mater. 2011, 23, 4504–4508

[ 1 ] J. A. Rogers , T. Someya , Y. Huang , Science 2010 , 327 , 1603 . [ 2 ] G. P. Collins , Sci. Am. 2004 , 291 , 74 . [ 3 ] K. Shiu , J. Zimmerman , H. Wang , S. R. Forrest , Appl. Phys. Lett.

2009 , 95 , 223503 . [ 4 ] S. R. Forrest , Nature 2004 , 428 , 911 . [ 5 ] M. B. Schubert , J. H. Werner , Mater. Today 2006 , 9 , 42 . [ 6 ] J. Engel , J. Chen , C. Liu , Appl. Phys. Lett. 2006 , 89 , 221907 . [ 7 ] S. Rosset , M. Niklaus , P. Dubois , H. Shea , J. Microelectromech. Syst.

2009 , 18 , 1300 . [ 8 ] D.-H. Kim , J. A. Rogers , Adv. Mater. 2008 , 20 , 4887 . [ 9 ] D.-H. Kim , N. Lu , R. Ghaffari , Y.-S. Kim , S. P. Lee , L. Xu , J. Wu ,

R.-H. Kim , J. Song , Z. Liu , J. Viventi , B. de Graff , B. Elolampi , M. Mansour , M. J. Slepian , S. Hwang , J. D. Moss , S. -M. Won , Y. Huang , B. Litt , J. A. Rogers , Nat. Mater. 2011 , 10 , 316 .

[ 10 ] R. J. Coffey , Artif. Organs 2008 , 33 , 208 . [ 11 ] J. S. Perlmutter , J. W. Mink , Annu. Rev. Neurosci. 2006 , 29 , 229 . [ 12 ] G. M. Whitesides , Nature 2006 , 442 , 368 . [ 13 ] I. M. Graz , D. P. J. Cotton , S. P. Lacour , Appl. Phys. Lett. 2009 , 94 ,

071902 . [ 14 ] S. Rosset , M. Niklaus , P. Dubois , H. R. Shea , Adv. Funct. Mater.

2009 , 19 , 470 . [ 15 ] G. Maggioni , A. Vomiero , S. Carturan , C. Scian , G. Mattei ,

M. Bazzan , C. de Julián Fernández , P. Mazzoldi , A. Quaranta , G. Della Mea , Appl. Phys. Lett. 2004 , 85 , 5712 .

[ 16 ] O. Akogwu , D. Kwabi , A. Munhutu , T. Tong , W. O. Soboyejo , J. Appl. Phys. 2010 , 108 , 123509 .

[ 17 ] D. S. Gray , J. Tien , C. S. Chen , Adv. Mater. 2004 , 16 , 393 . [ 18 ] J. Jones , S. P. Lacour , S. Wagner , Z. Suo , J. Vac. Sci. Technol. 2004 ,

22 , 1723 . [ 19 ] L. Ravagnan , G. Divitini , S. Rebasti , M. Marelli , P. Piseri , P. Milani ,

J. Phys. D: Appl. Phys. 2009 , 42 , 082002 . [ 20 ] A. L. Barabasi , H. E. Stanley , Fractal Concepts in Surface Growth ,

Cambridge University Press , Cambridge, UK 1995 . [ 21 ] P. D’Angelo , M. Barra , A. Cassinese , S. Guido , G. Tomaiuolo ,

Macro mol. Symp. 2007 , 247 , 43 . [ 22 ] F. Faupel , R. Willecke , A. Thran , Mater. Sci. Eng. R 1998 , 22 , 1 . [ 23 ] T. Adrega , S. P. Lacour , J. Micromech. Microeng. 2010 , 20 , 055025 . [ 24 ] E. Barborini , S. Vinati , M. Leccardi , P. Repetto , G. Bertolini ,

O. Rorato , L. Lorenzelli , M. Decarli , V. Guarnieri , C. Ducati , P. Milani , J. Micromech. Microeng. 2008 , 18 , 055015 .

[ 25 ] K. Wegner , P. Piseri , H. V. Tafreshi , P. Milani , J. Phys. D: Appl. Phys. 2006 , 39 , R439 .

In order to test the conductive performances of the samples during uniaxial strain cycles, the PDMS fi lm, which was prepared as previously described, was cut in rectangular 3 × 8 cm 2 stripes and attached to two microscope glass slides, with the 5 mm long central portion of the membrane suspended between the slides (see Figure 2 a). In the two extremities of the PDMS stripes, in the regions supported by the glass slides, two rectangular gold electrodes with resistances of ≈ 0.8 Ω were deposited by a standard evaporation process. The Au/PDMS nanocomposite fi lm was implanted by SCBI across the electrodes and through the suspended portion of the membrane (see Figure 2 a). The sample was then mounted on a custom-built motorized uniaxial stretcher by clamping only the glass-supported portion of the membrane to avoid the stretching of the gold-evaporated electrodes and to precisely defi ne the membrane’s portion subjected to strain as well as the percentage of strain. Clamping was also used to connect the evaporated electrodes to the contact pads, which, in turn, were connected to an Agilent 34410A multimeter. The stretcher and the multimeter were computer-controlled by LabVIEW software, allowing automatic acquisition of the nanocomposite electrical resistance with cyclic strain. The electrical resistance was recorded during measuring strain cycles at an elongation rate of 20 μ m s − 1 alternating with a series of fatigue strain cycles, at a faster elongation rate of 200 μ m s − 1 . Initially, subsequent measuring strain cycles were separated by a single fatigue strain cycle. As the total number of stretching cycles increased, the number of fatigue strain cycles between each measurement was progressively increased to 10, 50, 100, and 500.

Supporting Information Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements The authors thank Maura Francolini for her assistance with TEM characterization, Davide Marchesi and Gero Bongiorno for their assistance with SEM characterization, and Mary F. Ebeling for the syntactic revision of the manuscript.

Received: June 28, 2011 Revised: July 27, 2011

Published online: August 29, 2011

Journal of Physics D: Applied Physics

J. Phys. D: Appl. Phys. 47 (2014) 015301 (10pp) doi:10.1088/0022-3727/47/1/015301

Patterning of gold–polydimethylsiloxane(Au–PDMS) nanocomposites by supersoniccluster beam implantation

C Ghisleri1,2, F Borghi1, L Ravagnan2, A Podesta1, C Melis3, L Colombo3

and P Milani1,2

1 CIMAINA and Dipartimento di Fisica, Universita degli Studi di Milano, via Celoria 16, 20133 Milano,Italy2 WISE srl, Piazza Duse 2, 20122 Milano, Italy3 Dipartimento di Fisica, Universita di Cagliari, Cittadella Universitaria, 09042 Monserrato (Ca), Italy

E-mail: [email protected]

Received 19 August 2013, revised 10 October 2013Accepted for publication 24 October 2013Published 3 December 2013

AbstractPatterned gold–polydimethylsiloxane (Au–PDMS) nanocomposites were fabricated bysupersonic cluster beam implantation (SCBI) of neutral gold nanoparticles in PDMS throughstencil masks. The influence of nanoparticle dose on the surface roughness and morphology ofthe micropatterned regions of the nanocomposite was characterized. Nanoparticle implantationcauses the swelling of PDMS without affecting substantially the lateral resolution of the patterns.In order to have an insight on the mechanism and the influence of nanoparticle implantation onthe polymeric matrix, large-scale molecular dynamics simulations of the implantation processhave been performed. The simulations show that even a single cluster impact on PDMSsubstrate strongly affects the polymer local temperature and density. Our results show that SCBIis a promising methodology for the efficient fabrication of nanocomposite microstructures onpolymers with interesting morphological, structural and functional properties.

Keywords: nanocomposites, nanoparticles, stretchable polymers, micropatterning, AFM

(Some figures may appear in colour only in the online journal)

1. Introduction

Stretchable functional materials are enabling ingredients forthe fabrication of wearable electronics [1], smart prosthetics[2, 3] and soft robotics [4–6]. These applications require theintegration of electronic, optical and actuation capabilitieson soft, conformable and biocompatible polymeric substrates[7, 8]. In particular, the fabrication of stretchable patternedmicroelectrodes is necessary for actuation [6, 9], electricalstimulation and recording in neuroprosthetics [10–12].

Poly(dimethylsiloxane) (PDMS) is a very popularplayground for the proof-of-principle of soft devices sinceit couples biocompatibility with mechanical properties andmachinability suitable for the production of dielectricelastomeric actuators [13]. Incorporation of metalnanoparticles in PDMS thin films using reduction of chemicalprecursors by superficial penetration or direct incorporationof preformed clusters have been used for the fabrication of

lab-on-chip and optical devices ([14] and references therein).Particular efforts are also currently being concentrated on thefabrication of stretchable metallic circuits and microelectrodesintegrated on PDMS [9, 15, 16].

A straightforward approach to the metallization (e.g.with Au, Ag, Pt, Pd) of PDMS is metal vapour deposition(MVD) [17], however, the use of MVD is hampered by theweak adhesion of the metallic layer on the elastomer and thesubsequent delamination of the conducting layers even at verylow deformations [11, 18, 19]. The use of adhesion layers suchas Cr or Ti, or the treatment of the PDMS surface by oxygenplasma improve the performances, however the resilience isnot adequate for a large number of applications [18, 20].

An alternative to MVD is ion implantation where noblemetal ions are implanted with energies in the range ofkeV atom−1, thus forming a conductive layer just below thepolymeric substrate surface [21]; the diffusion of implantedions gives also rise to the formation of nanoparticles

0022-3727/14/015301+10$33.00 1 © 2014 IOP Publishing Ltd Printed in the UK

J. Phys. D: Appl. Phys. 47 (2014) 015301 C Ghisleri et al

Figure 1. (a) The SCBI apparatus. The mix of gas and clusters produced in a cluster source attached to the expansion chamber isaccelerated by a difference of pressure between the source and the expansion chamber and collimated by the aerodynamic focuser. Then thenanoparticles enter the deposition chamber and they are implanted in the polymeric substrate held on a movable sample holder allowing thedeposition on large areas though a rastering technique. (b) Scheme of the rastering process: by moving repeatedly the sample holdervertically and horizontally with respect to the cluster beam, it is possible to uniformly implant large samples.

via Ostwald-ripening phenomena [22, 23]. Stretchableelectrodes fabricated by ion implantation show adhesionand electrical conductivity degradation rates upon cyclicalstretching that are significantly better than evaporatedelectrodes, however a substantial modification of the polymericmatrix due to radiation-induced disruption of chemicalbonds, carbonization, bond reorganization and cross-linkingis observed [24–26]. The use of ions also results in the build-up of electrical charges within the polymer substrate in theinitial stages of the implantation process: this can perturbthe trajectories of the incoming ions, thus lowering the lateralresolution of patterning performed using shadow masks. Forthis reason, shadow masks can be used only for structureslarger than 100 µm; in general, micropatterning with highlateral resolution is obtained with ion implantation throughphotolithography steps or lift-off processes [21].

Recently we demonstrated that neutral metallic nanopar-ticles produced in the gas phase and aerodynamically accel-erated in a supersonic expansion can be implanted in a poly-meric substrate to form a conductive nanocomposite with supe-rior resilience and interesting structural and functional proper-ties [27, 28]. This approach is called supersonic cluster beamimplantation (SCBI) and it is based on the use of a highlycollimated supersonic beam carrying metallic clusters with akinetic energy of about 0.5 eV atom−1. Even if the kinetic en-ergy is significantly lower than in ion implantation, neutralclusters are able to penetrate up to tens of nanometres intothe polymeric target forming a conducting nanocomposite andavoiding electrical charging and carbonization [27, 28].

Supersonic cluster beams can be efficiently used forthe production of micrometre-scale patterns through stencilmasks [29, 30]. In particular, SCBI can produce micropatternsand microelectrodes on thin flexible polymeric substrates,such as SU8 using lift-off techniques [31]. Preliminaryresults showed also that electrically conducting regions can bepatterned on PDMS with SCBI using stencil masks [27]. High-resolution patterning achievable with SCBI is a considerableadvantage for soft devices microfabrication compared tochemical methods for functional nanocomposite production.

Here we present a characterization of 2D micropatternedregions of Au–PDMS nanocomposite integrated on PDMS bygold nanoparticle implantation through stencil masks. We havefocused our attention on the evolution of surface roughnessand morphology of Au–PDMS patterned nanocomposites andhow this affects the lateral resolution attainable with SCBI.In order to have an insight into the influence of nanoparticleimplantation on the polymeric matrix, we performed numericalsimulations of the implantation process.

2. Experimental section

2.1. Supersonic cluster beam implantation

The supersonic cluster beam used for the implantation wasproduced by a pulsed microplasma cluster source (PMCS), asdescribed in detail in [27, 32, 33]. Briefly, a PMCS consists ina ceramic body with a cavity where a metallic target (Au in thepresent case) is sputtered by a localized electrical dischargeignited during the pulsed injection of an inert carrier gas (Heor Ar) at high pressure (40 bar). The sputtered metal atomsfrom the target thermalize with the carrier gas and aggregatein the cavity forming metal clusters. The carrier gas–clustermixture expands out of the PMCS through a nozzle into alow pressure (10−6 mbar) expansion chamber (figure 1(a)).The supersonic expansion originating from the high pressuredifference between the PMCS and the expansion chamberresults in highly collimated supersonic beam: a divergencelower than 1◦ is obtained by using aerodynamic focusingnozzles [27, 33].

The central part of the cluster beam enters a secondvacuum chamber (deposition chamber, at a pressure ofabout 10−5 mbar) through a skimmer, and it impinges onthe polymeric substrate, supported by a motorized substrateholder. During implantation, the holder displaces the substratein the two directions orthogonal to the cluster beam axis,allowing implantation on an arbitrarily wide area with a highhomogeneity [31]. A typical scheme of this raster process issketched in figure 1(b).

2


Figure 2. (a) AFM topography map (2 µm × 1 µm) of gold nanoparticles on silicon substrate; (b) Au nanoparticles size distributionobtained from the AFM map.

The size distribution of the nanoparticles used forimplantation has been determined with atomic forcemicroscopy (AFM) [34], by imaging a sub-monolayer sampleobtained on a silicon substrate exposed to the nanoparticlebeam for 3 s at a deposition rate of 0.02 nm s−1. Figure 2(a)shows a representative topography of a 1 µm × 2 µm areaof gold nanoparticles deposited on Si. The diameter ofthe nanoparticles (Dz) has been determined as the heightof the objects. The distribution of heights (diameters) ofnanoparticles, calculated from the analysis of ten topographicimages, and the corresponding Gaussian fit are shown infigure 2(b). A mean value Dz = 3.9±1.4 nm (mean±standarddeviation) was obtained.

Supersonic expansion accelerates the clusters to a meanvelocity of approximately 1000 m s−1, meaning that the metalclusters are accelerated towards the polymeric substrate witha kinetic energy Ek of roughly 0.5 eV atom−1 [35]. Thisenergy is about four orders of magnitude less than the typicalkinetic energies for ion beam implantation in polymers [9, 20].Considering the atomic deposition rate Natm (number of atomsreaching surface unit area per second), one can calculate thesurface power density for the cluster implantation: Psurf =Natm ·Ek. This power density is of the order of some µW cm−2

in the case of SCBI, thus not producing a significant increase ofthe substrate temperature above room temperature (RT). Thisshould be compared with the power density for ion beamswhich is typically of the order of tenths of watts [21].

While for atom or nanoparticle deposition on a hardsubstrate one can define the thickness of the deposited material,in the case of nanoparticle implantation in polymers thisquantity is not well defined. During implantation we placeda rigid substrate (generally a half-masked silicon or glass)on the sample holder together with the polymeric substrateto be implanted so that both intercept the same sectionof nanoparticles beam. Since the amount of deposited orimplanted nanoparticles on the two substrates is the same,we can define the equivalent thickness teq of nanoparticlesimplanted in the nanocomposite as the thickness of the filmproduced by the same amount of nanoparticles deposited onthe hard substrate.

2.2. Elastomeric substrates

PDMS substrates used in this work were produced with aSylgard 184 Elastomer Kit (Dow Corning) by mixing for about

15 min the base and the curing agent in a 10 : 1 ratio. Themix was degased for 30 min in low vacuum in a desiccatorand casted in a 10 cm diameter Petri dish up to a thickness ofroughly 1 mm. After the polymerization at 100 ◦C for 1 h inan oven in ambient air, the PDMS was cut into pieces of thedesired dimensions.

2.3. Patterning of the Au–PDMS nanocomposites

We used stencil masks for the nanocomposite patterning withmicrometric resolution by exploiting the high collimationtypical of supersonic beams. Three transmission electronmicroscopy (TEM) grids B, C and D (respectively, G2760N,G2786N, G220-8 from Agar Scientific) and a small pieceof steel mesh (pattern E) were placed in front of the barePDMS substrate at a distance of roughly 500 µm from thesurface (figure 3(a)). The nickel grids have hexagonal, squareand round holes with micrometric sizes of 30 µm, 7.5 µmand 36.5 µm, respectively, while the steel mesh consists in4 µm × 90 µm and 60 µm spaced triangular apertures (seeinsets of figure 3(b)). Next to the masks a uniform region (areaA in figure 3(a), size of 5 mm × 5 mm) was homogeneouslyimplanted as a reference. Implantation through the masks wasperformed for 45 min at a rate of about 0.02 nm s−1, reachingan equivalent thickness of 59 nm.

2.4. Implantation of nanoparticles with a radial densitygradient

We characterized the effect of the dose of implantednanoparticles on the surface morphology of PDMS byproducing samples with a nanoparticles density gradient(figure 4). By moving the substrate during implantationonly in the vertical direction (as schematically shown infigure 4(a)) we obtained a homogeneous nanocomposite inthe direction parallel to the raster, with a gradient in thedose of nanoparticles implanted in the direction perpendicularto the raster. The deposition scheme of the sample withdensity gradient is shown in figure 4(a): a glass slide is halfcovered by a 13 mm wide and 76 mm long PDMS film, analuminum foil mask partially shadows the PDMS film and thebare glass substrate. This configuration allows producing inone step a sample where implanted regions can be comparedwith pristine PDMS and cluster-assembled film on glass for

3


Figure 3. (a) Scheme of the sample prepared for the morphological characterization. Three TEM grids (patterns B, C, D) and a steel texture(pattern E) were used as stencil masks. The area A in the bottom left corner of the sample serves as reference for a non-patternednanocomposite. (b) The four patterns in detail: the pictures represent the patterned PDMS observed with an optical microscope (50×magnification for pattern C, 20× magnification for the others), the insets in the top right corner in each picture represent the detailed schemeof the respective pattern.

the exact determination of the equivalent thickness gradient.The substrate underwent a SCBI process for 27 min with adeposition rate of 0.06 nm s−1, reaching a maximum equivalentthickness (in the central position of the gradient) of 100 nm(figure 4(b)). The equivalent thickness of gold nanoparticlesimplanted in the PDMS follows a symmetric bell-shape trendin the horizontal direction, as reported in figure 4(c).

2.5. AFM characterization

The surface morphology of the nanocomposites wasinvestigated using a Bioscope Catalyst/Nanoscope V AFM(Bruker Instruments). The AFM was operated in tapping modein air, using rigid cantilevers with resonance frequency 250–350 kHz, equipped with single crystal silicon tips with nominalradius 5–10 nm. In the case of the sample produced with theradial density gradient, several 2 µm × 1 µm (2048 × 512points) topographic maps were acquired on each of 12 differentregions distributed along the gradient of the deposition andseparated by about 100 µm from each other. The images wereflattened by line-by-line subtraction of first- and second-orderpolynomials in order to remove artefacts due to sample tilt andscanner bow. AFM topographies are plotted using a colourscale spanning a finite vertical height range (dark to bright).On each flattened AFM image the RMS roughness Rq wascalculated as the standard deviation of surface heights; Rq

values have been averaged and standard deviation of the meanhas been calculated as associated error.

2.6. TEM characterization

TEM was used for the evaluation of the clusters penetrationdepth dnc in PDMS. Small PDMS substrates (approximately5 mm long, 2 mm wide and 1 mm thick) were implanted withdifferent increasing doses in order to evaluate the evolution

of the implantation process and nanoparticle organization inthe matrix. We used implantation times of 3 s, 30 min and120 min at a rate of about 0.02 nm s−1, reaching an equivalentsub-monolayer on a rigid substrate and equivalent thicknessesof 40 nm and 140 nm, respectively. Ultra-thin slices (with athickness of about 100 nm) of nanocomposite were cut with acryo-ultramicrotome at −160 ◦C and laid down on TEM gridscovered with a layer of carbon-coated Formvar. Images wereacquired with a Philips CM10 (80 kV) microscope.

2.7. Wettability characterization

The wettability of the nanocomposite surface was analysedby contact angle measurements with First Ten AngstromsFTA200. A milli-Q water droplet (volume: 3 × 10−6 l)was deposited on the surface of area A of the patternednanocomposite sample and the contact angle value obtained byaveraging 150 measures from as many droplet images acquiredby a CCD camera immediately after the deposition.

2.8. Modelling

We performed large-scale molecular dynamics simulationsof the implantation process based on state-of-the-art forcefields [36]. While technical details of the present simulationshave been published elsewhere [37], here we remark on theirmost important features. First of all, in order to keep thepresent simulations on the relevant experimental length scalewe made use of very large-scale simulation cells containingas many as ∼4.6 million atoms. This system size correspondsto 25.6 × 25.6 × 85.35 nm3: indeed, a very large simulationcell (demanding a very intense computer effort), is necessaryif the full process of surface impact, the propagation intothe substrate and the stopping of the Au clusters must bereproduced. Such large systems have been aged for as many

4


Figure 4. (a) Scheme of the sample prepared for the morphologicalcharacterization of the nanocomposite as a function of theequivalent thickness. The rastering in the vertical direction allowsobtaining a gradient of nanoparticle dose implanted in the PDMSsubstrate. (b) Scheme of the sample after the implantation process:the deposition on glass serves as reference of a cluster-assembledfilm on a rigid substrate. (c) Equivalent thickness as a function ofthe horizontal position of the analysed sample.

as 180 ps. Such a relatively long simulation time was indeednecessary in order to allow the system to fully stabilize afterthe cluster impact.

Furthermore, special care was taken to model theinteractions between the implanted Au clusters and the targetsubstrate, as well as the possible presence of molecularlinkers into the PDMS film. Accurate benchmark calculationshave been executed on well-known physical properties ofpristine PDMS (such as density and radial distributionfunction), confirming the quantitative reliability of the presentcomputational set-up. Finally, the simulated protocol ofimplantation was chosen so as to mimic as close as possible theactual SCBI process; in particular, we implanted Au clusterswith radius of 3 nm with energies of 0.5, 1.0 and 2.0 eV atom−1.

Figure 5. TEM images of thin slices (about 100 nm thickness) of thePDMS implanted for 3 s (a), with an equivalent thickness of 40 nm(b) and 140 nm (c) of gold nanoparticles. The penetration depth ofthe nanoparticles is approximately 180 nm in each case.

3. Results and discussions

3.1. Cluster implantation depth

The implantation of gold nanoparticles for different teq hasbeen characterized by TEM imaging of ultra-thin slices ofnanocomposite cut by cryo-ultramicrotomy. Figure 5(a)shows the nanoparticles embedded in the polymer resultingfrom a supersonic beam exposure of the substrate of 3 s, theimplantation depth is roughly of 180 nm. A slight increase ofnanoparticles density towards the PDMS surface is visible: thiscan be explained in terms of the limited dynamic range of theCCD camera overexposing the polymer–vacuum interface. Byincreasing the dose of implanted nanoparticles to equivalentthicknesses of 40 nm (figure 5(b)) and 140 nm (figure 5(c)),an increase of the volume filling factor (defined as the ratiobetween the total volume of the nanoparticles and the volumeof the polymer in which they are implanted) is observed withoutany significant change in the cluster implantation depth. Theseobservations indicate that implantation depth does not dependon the dose of implanted nanoparticles. A further increasingof the dose results in the surfacing of the clusters on top of thepolymeric substrate and in the formation of a cluster-assembledmetallic layer on the PDMS surface [27].

3.2. Patterning and swelling

Different nanocomposite patterns were produced and analysed,as described in the experimental section: here we report theresults relative to hexagonal-shaped patterns; similar resultshave been obtained for the other patterns. Figure 6(a) showsthe AFM characterization of the pattern, figures 6(b) and(c) show an individual hexagon with a surface presentingfractures. The equivalent thickness of implanted nanoparticlesis teq = 59 ± 3 nm, while the measured height of Au/PDMShexagon is larger, being 100 ± 6 nm; this suggests that PDMS

5


Figure 6. (a) AFM morphological map of the hexagonal patterns; (b) single hexagonal pattern, z scale ranges from −100 to 300 nm; (c) 3Dmap of the same pattern; (d) topographic profile across the boundary between bare PDMS and Au-nc/PDMS indicated by the arrow in (b).

undergoes swelling upon cluster implantation. Swelling inPDMS is reported as a consequence of solvent exposure [38];far less studied is the effect of nanoparticle embedding on thereorganization of polymeric chains in silicones, although theformation of voids in polymers embedding inorganic particleshas been reported [39]. In our case swelling could degrade thelateral resolution obtainable with patterning through stencilmasks. To check this aspect we performed an AFM analysisof the boundary between bare PDMS and Au/PDMS hexagonalpattern (as indicated by the arrow in figure 6(b)). The effectivelateral resolution of the pattern ranges from 0.5 to 1 µmshowing that good lateral resolution can be obtained by SCBIin spite of the swelling phenomenon (figure 6(d)).

In order to clarify the relationship of the implantednanoparticle dose with the PDMS swelling, we systematicallyanalysed the evolution of the step between the bare PDMSsubstrate and the Au/PDMS nanocomposite in the samplewith a nanoparticle density gradient (see section 2.4). Theequivalent implantation thickness and the height of the swollenregion of Au/PDMS nanocomposite with respect to thebare PDMS substrate were calculated from the histogramof the heights of AFM images acquired across sharp stepsat the ns-Au/glass or Au-nc/PDMS boundaries, accordingly,produced by masking (figure 4(b)). The mean thickness/heightwas extracted as the distance between the peaks of the heightdistribution of the topographic maps; the errors associated withthe average quantities were calculated summing in quadraturea statistical error (the standard deviation of the mean of stepvalues calculated from different images), and a systematic error(7%), due to the non-linearity of the AFM piezo. The stepheight is always larger than the corresponding teq: the relation

Figure 7. Thickness of the swollen gold nanoparticles PDMSnanocomposite versus the equivalent thickness of implantedclusters. The linear equation fit of the data is also shown.

between teq and the swelling height is to a good approximationlinear as reported in figure 7. The swelling of PDMS growswith teq and even with teq = 100 nm it increases for more thana rigid offset without saturation in the investigated interval.

A possible origin of the observed swelling is themodification of the links between polymer chains caused by

6


Figure 8. (a) AFM topographic map of the Au-nc/PDMS, with adetail of a crack on the surface. (b) Phase map of the hexagonalpattern.

nanoparticle impact during implantation: in order to checkthe presence of partial break up and reorganization of links atthe polymer surface, we measured the wettability of implantedand pristine PDMS. For each of the two surfaces three series ofdata were acquired by depositing droplets in different points,in order to test the homogeneity of the sample. A contactangle of (113 ± 3)◦ was measured on the bare PDMS, whilethe PDMS implanted with an equivalent thickness of 59 nmof gold nanoparticles (area A in figure 3) is characterized bya contact angle of (112 ± 3)◦. These measurements suggestthat gold cluster implantation does not significantly affect themicroscopic structure of the PDMS surface.

Numerical simulations (see below) confirm that chainsbreakage is not occurring, whereas a local heating uponimplantation is produced; this might be the origin of a localdecrease of polymer density.

3.3. Surface morphology

Figure 6(b) and the 3D map in figure 6(c) show the presenceof scratches on the surface. In figure 8(a) a detail of oneof these surface defects, which appears like a fracture withelevated rims, is shown. The depth of the fracture as measuredfrom the AFM image is 25 nm, although due to the limitedpenetration of the AFM tip inside the high aspect ratio defectthis value must be taken as a lower limit for the actualfracture depth. We analysed the Au/PDMS roughness in 12

Figure 9. AFM topographical maps of PDMS (a), and threedifferent samples of Au-nc/PDMS with equivalent thicknesses∼20 nm (b), 40 nm (c) and 100 nm (d); z scale ranges from −20 to20 nm.

different positions along the gradient of deposited material,characterized by equivalent thickness teq ranging from 20 to100 nm. Figure 9 reports the evolution of surface morphologyupon the increase of the implantation dose from a pristinePDMS surface to three Au-nc/PDMS samples characterizedby increasing equivalent thicknesses (teq ≈ 20 nm, 40 nm and100 nm, respectively). Nanoparticles implantation in PDMSproduces an increasing corrugation as teq increases, with theappearance of craters and fractures which grow deeper andlarger as the Au cluster loading on PDMS increases. TheRMS roughness Rq of Au/PDMS nanocomposite evolves withteq: the quantitative analysis of the evolution of Rq with teq

is reported in log–log scale in figure 10. For comparisonpurposes, we report in figure 11 the AFM topography of aAu/PDMS nanocomposite and that of a nanostructured filmobtained by deposited gold cluster on a silicon substrate(ns-Au/Si) with the same equivalent thickness teq ≈ 20 nm.

7


Figure 10. Log–log plot of the roughness of the swollen goldnanoparticles PDMS nanocomposite versus the equivalent thickness.

Figure 11. AFM topographical maps of Au-nc/PDMS (a) andns-Au/Si (b). z scale ranges from −10 to 10 nm.

Table 1. RMS roughness Rq of the bare PDMS in different locationsof the patterned sample (see figure 3).

Sample Roughness Rq (nm)

Isolated PDMS 0.56 ± 0.01Pattern B 0.98 ± 0.01Pattern C 1.18 ± 0.01Pattern D 0.78 ± 0.03Pattern E 0.84 ± 0.01

Despite the similar thickness, roughness is markedly different:Rq of ns-Au/Si = 6.8 ± 0.1 nm, while Rq Au-nc/PDMS =3.9 ± 0.1 nm.

The implantation of gold clusters in PDMS induces achange of morphology also in the PDMS regions close to theimplanted area, as reported in table 1. The increase in PDMS

Figure 12. Temperature field (versus time and depth) of the EM-(left) and CL- (right) PDMS for the implantation energies of0.5 eV atom−1 (top), 1.0 eV atom−1 (middle) and 2.0 eV atom−1

(bottom).

roughness could be due to the horizontal swelling of the PDMSwith implanted Au clusters or to the diffusion of Au clustersin the PDMS polymeric substrate, even though in AFM phasemaps (figure 8(b)), a clear contrast between the bare PDMSand the Au-nc/PDMS nanocomposite is not visible.

3.4. Numerical simulation of cluster implantation

Neutral metallic cluster implantation in polymers hasbeen discovered only very recently [26, 27] and nodetailed physical models of this phenomenon are available.Computer experiments are currently underway to gain deeperunderstanding of the microscopic mechanisms and theinfluence of the structural and mechanical properties of thepolymeric matrix on the implantation. Recent simulations[36] provided evidence that for both entangled-melt (EM)and cross-linked (CL) PDMS the Au cluster penetration depthlinearly depends on the implantation energy, having an angularcoefficient of 7 nm eV−1 or 6 nm eV−1, respectively. Theseresults were explained in terms of the so-called clearing theway effect [40] which also predicts an inverse dependence ofthe same depth on the substrate cohesive energy.

In order to better characterize the implantation effect,we investigated the temperature field (wave) generated(propagating) inside the PDMS substrate upon cluster impact.In particular, we calculated the time-dependent temperatureprofile of the substrate during the implantation as a functionof the penetration depth from the PDMS surface. Resultsare shown in figure 12 for the EM- (left) and CL- (right)substrates, corresponding to implantation energies of 0.5 (top),1.0 (middle) and 2.0 (bottom) eV atom−1. For both EM-and CL-PDMS we observe a sudden temperature increase�TPDMS in the surface region: �TPDMS ∼ 20, 30 and

8


Figure 13. EM- (top) and CL- (bottom) PDMS surface height map for the implantation energies of 0.5 eV atom−1 (left), 1.0 eV atom−1

(centre) and 2.0 eV atom−1 (right).

50 K for increasing implantation energy. The correspondingtemperature wave generated upon the impact is dissipatedwithin the bulk substrate. After 180 ps we identify two maintemperature spots in the PDMS substrate, namely: a hot region(H) at T ∼ 320–350 K (figure 12, red–yellow shaded area)close to the surface and a cold region (figure 12, blue shadedarea) at T ∼ 300–310 K deeper into the substrate. Weremark that the overall temperature increase is always largerfor the cross-linked substrate, as a consequence of the differentstiffness: the deformation upon the impact is larger in the caseof the EM-PDMS substrate, thus resulting in a comparativelyreduced increase of temperature.

The extension of the H region depends also on theimplantation energy: by increasing its value from 0.5 to2.0 eV atom−1, we observe an increase of the H region depthfrom ∼10 nm up to ∼25 nm. Therefore, another effect ofa single cluster impact is to increase the PDMS surfacetemperature up to 50 K, thus generating a decrease of thesubstrate density as large as ∼5%. This feature is in goodagreement with the experimental finding of [41] that predicts aPDMS density decrease of ∼5% by increasing the temperaturefrom 300 to 350 K. Therefore, the next impinging clusters willpenetrate easily on a lower density PDMS substrate, furtherenhancing the effect of the local polymer density. This densitydecrease could explain the PDMS swelling experimentallyobserved upon the implantation.

We have extensively investigated the effect of a singlecluster impact on the PDMS surface morphology byperforming a theoretical analysis of the surface roughness. Byanalogy with the experimental AFM investigations reportedabove, we analysed the PDMS surface and internal nanoporesby means of a spherical Au probe of radius 1.0 nm placed ontop of the substrate. The probe interacts with the substrate viaa 9-6 Lennard-Jones potential having the ε and σ parameterstaken from the COMPASS force field. We spotted the PDMS

surface by a square grid of points with 0.5 nm spacing; at eachpoint of the grid we placed the probe and, by gently movingit up and down, we determined probe height correspondingto the minimum energy of the probe–substrate system. Theenvelope of such heights provided the simulated surfacetopography map, as shown in figure 13. The map, whichhas been suitably smoothed by means of a spline procedure,represents the surface height with respect to a conventionalzero set at the maximum atom vertical position in thePDMS film.

In all cases we notice the presence of a crater created on thesurface by the impinging Au cluster. The lateral dimensionsof the crater (as large as ∼6 nm) are almost unchanged byincreasing the implantation energy, while the crater depthstrongly depends on it. In fact, for EM-(CL-)PDMS, wemeasure crater depths of ∼6 (3) nm, ∼8 (6) nm and ∼10 (7) nmfor implantation energies of 0.5 eV atom−1, 1.0 eV atom−1

and 2.0 eV atom−1, respectively. The overall root-mean-square roughness (Rq) of EM-(CL-) PDMS increases with theimplantation energy: Rq = 1.1 ± 0.1 (0.9±1) nm, 1.8±0.1(1.7±0.1) nm and 2.5±0.1 (2.3±0.1) nm for implantationenergies of 0.5 eV atom−1, 1.0 eV atom−1 and 2.0 eV atom−1,respectively.

Shallower craters and an overall smaller roughness havebeen observed in the case of CL-PDMS. These results areconsistent with the reduced cluster penetration depth observedfor CL-PDMS [37]. This is due to the fact that in EM-PDMS the polymer chains are bonded only via dispersion andelectrostatic interactions, while in the CL-PDMS the chainsare partially covalently linked via the cross-linker molecule.Since the cluster penetration inside the PDMS matrix involvesthe breaking of the inter-chains bonds, we conclude thatduring the penetration inside the EM-PDMS substrate,the cluster experiences a weaker friction with respect toCL-PDMS.

9


4. Conclusions

We demonstrated the micrometric 2D patterning of Au–PDMSnanocomposite by supersonic cluster beam implantationthrough stencil masks on PDMS. The evolution of surfacemorphology and swelling of the patterned substrate canbe quantitatively correlated to the dose of implantednanoparticles. The PDMS swelling upon implantation doesnot substantially affect the lateral resolution of the pattern.

Atomistic simulations provide information on the PDMSmicrostructural evolution upon cluster implantation: theresults show that even a single cluster impact on the PDMSsubstrate remarkably changes the polymer local temperatureand density. Moreover, we observe the presence of craterscreated on the polymer surface having lateral dimensionscomparable to the cluster radius and depths strongly dependenton the implantation energy.

Our results suggest that SCBI is a promising methodologyfor the efficient and easy fabrication of nanocompositemicrostructures on polymers with interesting morphological,structural and functional properties.

Acknowledgments

The authors acknowledge Regione Lombardia and RegioneSardegna for their financial support to the project ‘ELDABI- Elettronica Deformabile per Applicazioni Biomediche’(project n. 26599138). We also acknowledge computationalsupport by CINECA (Bologna, Italy) under project ISCRA-UCIP, Maura Francolini, Elisa Sogne and Fondazione Filaretefor support in TEM and Contact Angle measurements

References

[1] Pang C, Lee C and Su K-Y 2013 J. Appl. Polym. Sci. 130 1429[2] Moon J-H, Baek D H, Choi Y Y, Lee K H, Kim H C and

Lee S-H 2010 J. Micromech. Microeng. 20 025032[3] Baek J-Y, An J-H, Choi J-M, Park K-S and Lee S-H 2008

Sensors Actuators A 143 423[4] Nawroth J C, Lee H, Feinberg A W, Ripplinger C M,

McCain M L, Grosberg A, Dabiri J O and Parker K K 2012Nature Biotechnol. 30 792

[5] Buselli E, Smith A, Grover L, Levi A, Allman R, Menciassi A,Mattoli V and Beccai L 2011 Microelectron. Eng. 88 1676

[6] Kovacs G, Lochmatter P and Wissler M 2007 Smart Mater.Struct. 16 S306

[7] Ma Z 2011 Science 333 830[8] Rogers J A, Someya T and Huang Y 2010 Science 327 1603[9] Rosset S and Shea H R 2013 Appl. Phys. A 110 281

[10] Cyganowski A, Minev I R, Vachicouras N, Musick K andLacour S P 2012 IEEE Sensors 2012 Conf. (28–31 October)paper 1974

[11] Graudejus O, Morrison B, Goletiani C, Yu Z and Wagner S2012 Adv. Funct. Mater. 22 640

[12] Lacour S P, Benmerah S, Tarte E, FitzGerald J, Serra J,McMahon S, Fawcett J, Graudejus O, Yu Z and Morrison BIII 2010 Med. Biol. Eng. Comput. 48 945

[13] Molberg M, Leterrier Y, Plummer C J G, Walder C, Lowe C,Opris D M, Nuesch F A, Bauer S and Månson J-A E 2009J. Appl. Phys. 106 054112

[14] Berry K R Jr, Russell A G, Blake P A and Roper D K 2012Nanotechnology 23 375703

[15] Zhang Y, Xu S, Fu H, Lee J, Su J, Hwang K-C, Rogers J A andHuang Y 2013 Soft Matter 9 8062

[16] Adrega T and Lacour S P 2010 J. Micromech. Microeng.20 055025

[17] Graz I M, Cotton D P J and Lacour S P 2009 Appl. Phys. Lett.94 071902

[18] Douville N J, Li Z, Takayama S and Thouless M D 2011 SoftMatter 7 6493

[19] Tuinea-Bobe C L, Lemoine P, Manzoor M U, Tweedie M,D’Sa R A, Gehin C and Wallace E 2011 J. Micromech.Microeng. 21 115010

[20] Akogwu O, Kwabi D, Munhutu A, Tong T and Soboyejo W O2010 J. Appl. Phys. 108 123509

[21] Rosset S, Niklaus M, Dubois P and Shea H 2009 Adv. Funct.Mater. 19 470

[22] Popok V N 2012 Rev. Adv. Mater. Sci. 30 1[23] Bechelany M, Maeder X, Riesterer J, Hankache J, Lerose D,

Christiansen S, Michler J and Philippe L 2010 Cryst.Growth Des. 10 587

[24] Jagielski J, Piatkowska A, Aubert P, Thome L, Turos A andAbdul Kader A 2006 Surf. Technol. 200 6355

[25] Abdul-Kader A M, Turos A, Grambole D, Jagielski J,Piatkowska A, Madi N K and Al-Maadeed M 2005 Nucl.Instrum. Methods B 240 152

[26] Clough R L 2001 Nucl. Instrum. Methods B 185 8[27] Corbelli G, Ghisleri C, Marelli M, Milani P and Ravagnan L

2011 Adv. Mater. 23 4504[28] Ravagnan L, Divitini G, Rebasti S, Marelli M, Piseri P and

Milani P 2009 J. Phys. D: Appl. Phys. 42 082002[29] Barborini E, Piseri P, Podesta A and Milani P 2000 Appl. Phys.

Lett. 77 1059[30] Barborini E et al 2008 J. Micromech. Microeng.

18 055015[31] Marelli M, Divitini G, Collini C, Ravagnan L, Corbelli G,

Ghisleri C, Gianfelice A, Lenardi C, Milani P andLorenzelli L 2011 J. Micromech. Microeng. 21 045013

[32] Barborini E, Piseri P and Milani P 1999 J. Phys. D: Appl.Phys. 32 L105

[33] Wegner K, Piseri P, Tafreshi H V and Milani P 2006 J. Phys.D: Appl. Phys. 39 R439

[34] Keller D 1991 Surf. Sci. 253 353[35] Milani P and Iannotta S 1999 Cluster Beam Synthesis of

Nanostructured Materials (Berlin: Springer)[36] Sun H 1998 J. Phys. Chem. B 102 7338[37] Cardia R, Melis C and Colombo L 2013 J. Appl. Phys.

113 224307[38] Lee J N, Park C and Whitesides G M 2003 Anal. Chem.

75 6544[39] Shirazi Y, Ghadimi A and Mohammadi T 2012 J. Appl. Polym.

Sci. 124 2871[40] Yamamura Y and Muramoto T 1992 Phys. Rev. Lett.

69 1463[41] Shih H and Flory P J 1972 Macromolecules 56 758

10

Part II

Reflective stretchable optics

CHAPTER 4

Fabrication and characterization of elastomeric opticaldevices

Elastomeric optical devices fabricated by Supersonic Cluster Beam Implantation of silvernanoparticles in PDMS were optically and morphologically characterized. In particularthe performances of elastomeric nanocomposite-based mirrors and diffraction gratingswere studied and compared with the state of the art.

4.1 Fabrication of the devices

Polydimethilsiloxane (PDMS) represents the ideal substrate for the fabrication of elas-tomeric optical devices, in terms of easy of processing, low cost, wide availability andgood mechanical properties [12].

The first step of elastomeric mirrors and diffraction gratings fabrication process con-sists in the production of the PDMS substrate, following two approaches according tothe wanted device. For the fabrication of elastomeric mirrors liquid PDMS was simplyspin-coated on a glass slide, obtaining a uniform thin polymeric film. The surface of acommercially available rigid diffraction grating was replicated with PDMS by moldingtechnique [12, 23] for the fabrication of elastomeric diffraction gratings.

PDMS substrates were implanted by SCBI of Ag nanoparticles with different equiva-lent thicknesses, up to 290nm, in order to study the effect of the polymer filler loading onthe optical properties of the nanocomposite. Silver was chosen for its low cost and su-perior and uniform reflectivity in the visible range compared to other metallic materials(gold, for example).

Similar PDMS substrates were metallized by means of thermal evaporation for acomparison of the optical and morphological properties with the nanocomposite-baseddevices.

4.2 Morphological characterization

In the previous chapter the surface roughness was investigated in order to control thefraction of impinging light scattered from the device surface. Here the morphologicalproperties relative to the devices were characterized, like the cracking of the reflectivelayer and the profile of the metallized elastomeric diffraction grating.

The superior resilience of the nanocomposite reflective layer undergoing mechanicalstress was demonstrated by investigating the surface of SCBI and evaporated mirrorsand gratings with an optical microscope in different conditions: as deposited, after peel-ing (i.e. removal of the nanocomposite from the supporting substrate) and after thou-sands of 50% stretching cycles. Evaporated samples rapidly encounter a deterioration

59

60 4.3 Optical properties

due to the breakage of the rigid conductive layer on top of the polymer surface, whileSCB implanted devices do not show significant changes in their surface morphology.

The different effects of the reflective layer behavior upon mechanical stress are par-ticularly important in the case of diffraction gratings: here the formation of cracks, withdimensions and periodicity comparable with the grooves of the grating, strongly affectthe performances of the optical device. For this purpose, the morphology of the elas-tomeric gratings was further studied by acquiring AFM single-line scans of both thebare and metallized grating profiles. The results show a very good agreement betweenthe profiles of the bare PDMS and the SCBI grating (also thanks to the high collimationof the cluster beam, allowing to perfectly reproduce the three dimensional structure ofthe grating), while a highly irregular profile is observed in the case of the evaporatedgrating.

4.3 Optical properties

4.3.1 Reflectance

Reflectivity of the nanocomposite superficial layer represents the most important opticalproperty of the devices studied in this work. Samples with different equivalent thick-nesses (30nm, 60nm, 120nm and 290nm) were analyzed in order to understand the effectof different nanoparticles concentrations on the trend of the surface reflectivity. Theequivalent thickness maximizing the reflectivity of the device must be determined.

The optical setup for reflectance spectra measurement, shown in Figure4.1a, consistsin a halogen lamp as light source, positioned at a large distance from a diaphragm (ide-ally at infinite distance) aiming to filter the white light generated by the lamp. The sam-ple, held on a tilting stage for its fine alignment with the detector, immediately followsthe diaphragm. The white light beam impinges on the nanocomposite with an angle of45◦ and the reflected beam is collected by a computer controlled monochromator. Themonochromator is equipped with a 600 lines/mm grating and a CCD camera (controlledwith the monochromator) for the spectra acquisition in the vis-NIR range (350-1100nm).Reflectance R(ω) was calculated as the ratio between the intensity IR(ω) of the light re-flected by the sample and the intensity I0(ω) of impinging light:

R (ω) =IR (ω)

I0 (ω)(4.1)

In order to acquire the spectrum of the impinging light while maintaining the opticalsetup unaltered, I0 was measured by replacing the sample with a commercial mirror.

The results, shown in Figure4.1b, present an interesting behavior. Reflectance of the30nm-implanted PDMS is very poor, around 20% throughout the whole visible range,as expected since the low number of implanted particles. Reflected light intensity isstrongly enhanced by the increase the equivalent thickness. Reflectivities between 60%and 70% are achieved for an equivalent thickness of 60nm, however reflected light inten-sity suddenly drops to 30%-40% for equivalent thicknesses of 120nm and 290nm respec-tively. This deterioration of the reflectivity for high nanoparticles concentration in thepolymer depends on the complex behavior of the surface plasmon resonance (extensivelystudied in chapter 7) and, for even higher equivalent thicknesses, to the cracking of thecluster layer growing on the polymer surface once the implanted layer of PDMS is satu-rated. As a result of this characterization, an equivalent thickness of 60nm was chosen asa standard for the fabrication of all the samples and devices studied in this work (exceptwhere otherwise specified).

Fabrication and characterization of elastomeric optical devices 61

Figure 4.1: (a) Optical setup for the acquisition of reflectance spectra. (b) Reflectance spectra of theAg/PDMS nanocomposite with different implanted equivalent thicknesses. The best reflectivityis achieved with an equivalent thickness of 60nm.

Reflectance of 60nm SCB implanted nanocomposite mirror was compared with asilver-evaporated PDMS mirror and the state of the art in the field of tunable gratings,consisting in gold-evaporated PDMS thin films [82]. Silver evaporated mirrors presenta reflectivity similar to silver implanted PDMS while gold-evaporated PDMS show amuch lower reflectivity (between 20% and 40%), dramatically dropping after few me-chanical deformations.

4.3.2 Spot size and linearity

The effect of the irregular profile of the evaporated grating was quantitatively studiedby measuring the diffracted spot size (referred as the solid angle subtended by the spot)when illuminated with monochromatic line (He-Ne laser, wavelength 632.8nm). Unlikenanocomposite-based gratings, in which the solid angle initially decreases and then re-main stable up to 50% stretching, a strong instability of the solid angle is observed forgratings metallized by evaporation. Moreover, the formation of large cracks perpendic-ular to the grating grooves on the surface of the evaporated grating gives rise to the ap-

62 4.4 Focusing properties

pearance of an unexpected and unwanted diffraction pattern orthogonal to the expectedone.

The linear behavior of the diffraction angle as a function of the stretching percentagewas tested too with good results for both the evaporated and SCBI gratings, leading tothe fabrication of stretchable-based spectrometers, as deeper explained in the next chapter.

4.4 Focusing properties

The high conformability and extremely low surface roughness of the obtained nanocomposite-based optical devices suggest their application to non-optical grade curved surfaces inorder to add optical power to the diffracting one. This property is requested in a num-ber of optical mounts [63] for the correction of aberrations like astigmatism or higherorder aberrations. In particular a SCBI elastomeric grating was applied to a roughly pol-ished cylindrical surface and the focal length of the curved device measured both on thereflected and on the diffracted beams. Results fully compatible with the expected val-ues were obtained, opening the way to the application of elastomeric optical devices onarbitrarily shaped non-optical grade surfaces for the correction of higher-order imagesaberrations. Several imaging applications may take advantage of this superior capabil-ity of nanocomposite-based diffraction gratings, hyperspectral imaging among them, asextensively discussed in the Conclusions and Perspectives.

LASER & PHOTONICSREVIEWS Laser Photonics Rev. 7, No. 6, 1020–1026 (2013) / DOI 10.1002/lpor.201300078

OR

IGIN

AL

PAPE

R

Abstract Stretchable and conformable optical devices openup very exciting perspectives for the fabrication of systems in-corporating diffracting and optical power in a single element.Supersonic cluster beam implantation of silver nanoparticles inan elastomeric substrate grooved by molding allows effectivefabrication of cheap and simple stretchable optical elementsable to withstand thousands of deformations and stretching cy-cles without any degradation of their optical properties. Thenanocomposite-based reflective optical devices were charac-terized both morphologically and optically showing excellentperformances and stability compared to similar devices fabri-cated with standard techniques. The nanocomposite-based de-vices can therefore be applied to arbitrary curved nonopticalgrade surfaces in order to achieve optical power and to min-imize aberrations like astigmatism. The high resilience of thenanocomposite material on which the devices are based allowsthem to be peeled and reused multiple times.

Nanocomposite-based stretchable optics

Cristian Ghisleri1,2, Mirko Siano2, Luca Ravagnan1, Marco Alberto Carlo Potenza2,and Paolo Milani1,2,∗

1. Introduction

Adaptive optics is a technology based on optical systemsthat can dynamically change their shape to compensate foroptical artifacts (due to optical aberrations, for example)introduced by the medium between the object and the im-age [1]. In biological systems (like the human eye) this isobtained by the capability of mechanically changing con-formation to “accommodate” for the changes of the opticalconditions.

In view of a large number of applications in spec-troscopy, microscopy [2, 3], optical telecommunications[4], aerospace [5] and in vivo medical imaging [6, 7], avery strong interest is focused on the fabrication of opti-cal elements such as lenses [2, 8], gratings [9–11], mirrors[3, 4, 6, 7, 12], able to change their geometrical shapes thusimproving the image quality. Different alternatives havebeen proposed for the replacement of rigid optical ele-ments (in particular gratings and mirrors) fabricated onglass or metals with adaptive elements such as electrostati-cally actuated suspended ribbons forming grating surfaces[13], complex comb-driven diffraction gratings [14] andpiezoelectric-driven tunable gratings [15]. These technicalapproaches are finding applications, however, they are stillconsidered too expensive and technologically complex toallow for a widespread use of adaptive optics solutions [16].

1 WISE s.r.l., Piazza Duse 2, 20122 Milano, Italy2 CIMAINA, Physics Department, Universita degli Studi di Milano, Via Celoria 16, 20133 Milano, ItalyThis work is dedicated to the memory of Gabriele Corbelli.∗Corresponding author: e-mail: [email protected]

Elastomeric substrates could represent simple, low-costand effective alternatives for the fabrication of optical el-ements that can be stretched and deformed. Transparentstretchable optical components operating in transmissionmode and made of silicone have been demonstrated forthe fabrication of lenses [17], light valves [18], transmis-sion diffraction gratings and wave-front engineering de-vices [19, 20]. Since these devices work in transmission,light passing through the polymer undergoes absorption,refraction and scattering due to defects in the polymericmatrix.

Despite several attempts, the fabrication of reflectivestretchable optical elements consisting of an elastomericsubstrate covered by a reflecting metal film is affectedby several problems: the metallization of elastomeric sub-strates by classical thin-film coating techniques (e.g. ther-mal metal evaporation, e-beam metal evaporation, elec-trodeposition, etc.) results in a poorly adherent reflectinglayer undergoing cracking and delamination, even uponvery small deformations [17, 21–24]. The mismatch be-tween the metal and the supporting elastomer mechanicalproperties causes, also in static conditions, the formation ofbuckling instabilities and wrinkles on the surface [25–27].

Here, we demonstrate the fabrication of stretchable re-flective gratings based on metal–elastomer nanocompositesobtained by supersonic cluster beam implantation (SCBI) in

C© 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

ORIGINALPAPER

Laser Photonics Rev. 7, No. 6 (2013) 1021

Figure 1 (a) Fabrication process of thePDMS grating replica from a commercialdiffraction grating master and subsequentmetallization. (b) Picture of the stretchableand deformable reflective diffraction grating.(c) PDMS grating withstanding arbitrary de-formations without deterioration. (d), (e), (f)AFM single profiles of the bare PDMS grat-ing, the 60-nm Ag SCBI and the 60-nm Agthermally evaporated gratings, respectively.

polydimethylsiloxane (PDMS). The implantation of elec-trically neutral metallic nanoparticles accelerated in a su-personic expansion is a very effective method to metallizea polymer surface with stable and resilient layers [28]. Al-though the kinetic energy per atom in supersonically accel-erated neutral clusters is four orders of magnitude lowerthan in the case of metal ion implantation in polymers[29], clusters (made of several thousand atoms) have suffi-cient inertia to penetrate inside the polymeric target and toform a nanocomposite layer, while avoiding charging andcarbonization of the polymeric substrate [26, 30]. Metal-lic electrodes and micropatterns on stretchable substratesdeposited by SCBI capable to sustain very large deforma-tions with improving electrical conductance with cyclicaldeformation have been recently reported [28].

We show that stretchable diffraction gratings obtainedby SCBI can easily withstand elongations up to 50% forthousands of cycles with no deterioration of their opticalquality and dispersion performances. Moreover, their su-perior flexibility and resilience allow fitting the gratingsupon surfaces of a given shape, thus making very simpleand cheap imposing optical power in addition to dispersionproperties, as it is currently needed in a number of opticaldevices [31–35].

2. Grating fabrication and morphologicalcharacterization

Transparent gratings were fabricated by making a PDMSreplica from a master consisting of a commercial glass holo-graphic diffraction grating (1200 grooves mm−1); the mold-ing procedure is schematically shown in Fig. 1a. PDMSliquid precursor (Sylgard 184 Silicone Elastomer Kit fromDow Corning, mixed in a 10:1 ratio) was cast on the glassdiffraction grating and then crosslinked at RT for 48 h.After crosslinking the mold was carefully detached from

the master, resulting in a grating with dimensions of about25 mm × 12 mm and about 1 mm thick.

The transparent PDMS grating was then made reflec-tive by implanting silver nanoparticles by means of a super-sonic cluster beam implantation (SCBI) apparatus equippedwith a pulsed microplasma cluster source (PMCS) [36,37].PMCS consists in a ceramic body with a cavity in whicha solid silver target (purity 99.9%) is vaporized by a lo-calized electrical discharge supported by a pulsed injectionof an inert gas (He or Ar) at high pressure (40 bar). Themetal atoms, sputtered from the target, aggregate in thesource cavity to form metal clusters; the mixture of clustersand inert gas expands through a nozzle forming a super-sonic beam into an expansion chamber kept at a pressureof about 10−6 mbar. Electrically neutral nanoparticles exit-ing the PMCS are aerodynamically accelerated in a highlycollimated beam with divergence lower than 1◦ and witha kinetic energy of roughly 0.5 eV atom−1 [28]. The cen-tral part of the cluster beam enters, through a skimmer, asecond vacuum chamber (deposition chamber) where thebeam is intercepted by the polymeric substrate. A raster ofthe nanoparticle beam on the PDMS grating is obtained bya computer-controlled motorized substrate holder duringthe implantation process. The measure of the quantity ofmaterial implanted into the PDMS is obtained by placinga half-masked hard substrate (glass or silicon) next to thegrating, during the implantation process, in order the inter-cept the same number of nanoparticles. The thickness ofthe cluster-assembled film deposited on the hard substratecorresponds to the equivalent thickness teq in the case ofthe metal–polymer nanocomposite. We metallized differ-ent PDMS gratings and plane PDMS substrates with differ-ent equivalent thicknesses, up to 160 nm, with a depositionrate of about 0.08 A s−1. The deposition rate was calculatedas the ratio between the final equivalent thickness and theoverall implantation time on an area of 750 mm2 requiredfor a homogeneous implantation of the substrate.

www.lpr-journal.org C© 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

LASER & PHOTONICSREVIEWS

1022 C. Ghisleri et al.: Stretchable optics

A reflective grating with teq of 60 nm is shown in Fig. 1b:it can be stretched and deformed without altering the macro-scopic optical quality of the surfaces (Fig. 1c). Figure S1in the Supporting Information Section shows a scanningelectron microscopy image of the nanocomposite gratingsurface. The surface roughness of the nanocomposite ismuch smaller than the pitch of the grating and the lightwavelength in the visible range, thus not affecting the opti-cal properties of the device.

In order to compare the performance of the nanocom-posite grating with devices obtained with a traditional ap-proach, we metallized PDMS gratings with silver and goldfilms at different thicknesses using a standard thermal evap-oration method. In particular, we produced PDMS gratingswith 26-nm and 60-nm thick silver and 26-nm thick goldmetallic layers with a deposition rate of about 0.6 A s−1.We characterized the diffracting side of bare and metallizedPDMS gratings by atomic force microscopy (AFM): Fig. 1dshows an AFM single line scan of the profile of the barePDMS grating, presenting a sinusoidal shape with a peak-to-valley height of about 50 nm and a pitch of 833 nm,exactly reproducing the features of the master, having agroove density value of 1200 grooves mm−1. Figure 1e re-ports the same single line scan taken on the PDMS gratingmetallized by SCBI: an excellent reproduction of the bareprofile is observed, while the profile relative to the 60 nmAg thermally evaporated PDMS grating (Fig. 1f) showsa more irregular profile compared to the nanocompositegrating. The deposition rate used for classical metallizationproduced surfaces with a better morphology (see Figs. S2and S3 of the Supporting Information Section) comparedto what has been reported in the literature [21,26,27], how-ever, the wrinkling of the surface has dimensions compara-ble with the grating pitch.

The difference in the peak-to-valley height between thetwo metallization techniques may be partly due to the highdivergence of the thermally evaporated metal atoms caus-ing a shadowing effect and hence a nonuniform coating ofpeaks and valleys. The very low divergence of nanoparticlesaccelerated in a supersonic beam results in a high lateralresolution of the implanted layer, allowing the very precisereproduction of submicrometer grooves [38, 39].

3. Optical characterization

In order to characterize the optical performances of thegratings, we performed different experiments, aimed to testi) the reflectance of the nanocomposite ii) the grating re-sponse under controlled uniaxial stretching. For the opticalcharacterization three silver SCBI and two silver evapo-rated gratings (respectively named C1, C2, C3 and E1, E2)were tested.

The reflectance of the silver–PDMS nanocomposite wasmeasured on plane, mirror-like, silver implanted samples. A45◦ incident white-light beam from a halogen lamp was re-flected by the sample into a monochromator (coupled with aCCD camera for the spectral acquisition), while a standard

Figure 2 VIS-NIR reflectance (350–1100 nm) with 45◦ incidentwhite light for the 60-nm Ag SCB Implanted (red line) and the26-nm Ag evaporated (blue line) PDMS mirrors after peeling andstretching. The lower green lines represent the reflectance of a26-nm Au evaporated PDMS mirror before (dashed) and afterpeeling and stretching (solid).

mirror was used as reference. Figure 2 shows the visible-NIR reflectance of a SCBI silver nanocomposite (teq =60 nm) compared with a thermally evaporated silver-PDMSmirror (t = 30 nm) and gold-PDMS mirror (t = 26 nm).A teq of 60 nm provides the best reflectance among allthe tested nanocomposites with different equivalent thick-nesses. Larger amounts of implanted nanoparticles lead tothe increase of light absorption due to surface plasmonresonance, thus modifying the reflectance spectrum of thesurface [40,41]. By further increasing the equivalent thick-ness, nanoparticles saturate the polymer and begin to growon the surface of the polymer, forming a thick layer under-going larger cracking and thus causing a loss of reflectanceof the device.

We also characterized stretchable devices obtained bygold evaporation since in the literature the use of goldcoating has been reported for the fabrication of tunablediffraction gratings [42]. All the mirrors underwent severalmechanical stresses due to manipulation and peeling fromthe sample holder prior to the measurement, followed by3000 cycles of uniaxial stretching at 50% of elongation.The implanted and evaporated silver mirrors (blue and redlines) exhibit a similar reflectance, not undergoing signifi-cant modifications after stretching, whereas the gold mirrorshowed a severe deterioration (green solid line), due to thepoor adhesion of gold on PDMS. In Fig. S2 we reportedoptical microscope images of a 26-nm thick gold film de-posited on a flat PDMS surface and on a PDMS grating.The peeling of the gold evaporated polymeric film fromthe sample holder is sufficient to cause deterioration of the

C© 2013 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.lpr-journal.org

ORIGINALPAPER


Figure 3 (a) Schematics of the optical layout used to characterize the stretched gratings. (b) and (c) show results obtained bystretching 60-nm Ag evaporated and 60-nm Ag SCBI gratings, respectively, after 10 000 stretching cycles. Horizontal scales showthe stretching imposed to the gratings (%). Left vertical scale represents the elongation of the grooves spacing; right vertical scalerepresents the solid angle subtended by the diffracted spot. (d) and (e) show the surfaces of the evaporated and the SCBI gratingsrespectively after ten thousand stretching cycles, as seen at the optical microscope (20× magnification). (f) Diffraction pattern of thesilver evaporated elastomeric grating. The two spots in the horizontal direction are due to the grooves of the grating, while the unwantedvertical spots (the symmetric spot is present but hidden by the mirror mount) are due to the uncontrolled cracking of the superficialmetal layer of the grating. These vertical spots are not present in the case of the SCBI grating.

surface, while after 3000 cycles at 50% stretching the sur-face is completely covered by cracks.

For the characterization of the behavior under uniaxialstretching we only considered silver implanted (C2, teq =60 nm) and evaporated (E2, t = 60 nm) reflective gratings.The gratings were clamped on a custom-built motorizeduniaxial stretcher with the periodic structure of the grat-ing perpendicular to the stretching direction. A monochro-

matic light beam (coming from a He-Ne laser, wavelengthλ = 632.8 nm, power P = 1 mW) is sent through a cu-bic beam splitter, normally to the entrance face. The 90◦reflected beam impinges normally onto the center of thefree-standing clamped elastomeric grating (Fig. 3a). Theposition l of the spot of the 1st-order diffracted beam re-spect to the one relative to the 0th order on a screen ata distance d from the grating is related to the diffraction




angle θ through the following equation, obtained from sim-ple geometrical considerations:

θ = arcsin

(l

d

)(1)

Both l and d were manually measured with a precisionruler, the distances being of the order of 100 mm. The elas-tomeric grating was stretched stepwise up to a 25% maxi-mum extension, and for each step the corresponding valueof l has been determined. Both the size and shape of thediffracted spot at the screen have also been characterized.

An example of the experimental results after 10 000thousands stretching cycles for E2 and C2 is shown inFigs. 3b and c. In both cases (evaporated and SCBI) theyshow the expected linearity of the grating response as afunction of the stretching. Besides, the solid angle sub-tended by the diffracted spot at the screen is reported foreach case, showing a good stability overall the elongationrange for the nanocomposite grating, while the evaporatedone shows large fluctuations. In both cases, a small elonga-tion is needed to reduce the spread of the diffracted light atrest position. Similar results have been obtained for all thetested gratings.

The superior resilience of SCBI gratings compared toevaporated ones is evident after thousands of stretchingcycles. The optical microscope image of the surface ofthe evaporated grating after ten thousand stretching cycles(Fig. 3d) shows an evident cracking of the rigid metal layerboth in the stretching and the orthogonal directions, causingthe rapid increase of stray light and the arise of unexpecteddiffraction beams perpendicular to the stretching direction(Fig. 3f). The surface of the SCBI grating (Fig. 3e) presentsonly minor cracks, not affecting the optical quality of thedevice. The grating mounted on the uniaxial stretcher canbe exploited to fabricate a very simple and cheap “linear”alternative to traditional rotating grating monochromators[31] as shown in the Supporting Information (Fig. S4 andthe videoclip).

After a few weeks from fabrication both the implantedand the evaporated silver gratings underwent deteriorationdue to silver oxidation. In order to avoid this problem wedeposited a very thin PDMS layer (about 10 μm thick) byspin coating on the metallized gratings in order to reduceoxygen exposure. The use of a capping layer is detrimentalto the optical quality of the evaporated silver grating, evenat very low metal thicknesses, as we report in Fig. S3 in theSupporting Information.

The stretchability and good resilience of the silvernanocomposite gratings allow them to be fitted to curvedsurfaces, thus adding optical power to the diffracting ele-ment. The following test, regarding the application of thestretchable grating to a cylindrical surface, is a benchmarkfor a number of possible applications in which the grat-ings are used to modify the wavefronts of the emergingbeam, both in traditional and adaptive optics applications[31, 32]. For example, this can be used to compensate foraberrations, to limit the number of optical elements within

a device and to cut down the fabrication costs of concaveor arbitrary shaped gratings, currently used in a numberof grating mounts [31]. The breakthrough here is repre-sented by the separate production of identical conformablediffractive elements with an optical-grade surface to be sub-sequently adapted to rigid substrates with different shapesand curvatures. This can revolutionize the fabrication ofdifferent optical systems: producing arbitrarily curved andnonoptical grade shaping surfaces with traditional or com-puter numerical control (CNC) machines is much easier andcheaper than blazing a curved optical-grade surface [31].

For the proof-of-principle of this approach, we havechosen cylindrical surfaces in order to preserve the spac-ing between the grooves over all the grating extension. Itshould be noted that this is rigorously true regardless of thegrooves’ orientation over the surface, thanks to the perfectEuclidean geometry of cylindrical surfaces. The astigma-tism of the system can be exploited either to compensatefor aberrations, or in connection with other cylindrical sur-faces with optical power in the orthogonal direction to givea spherical shape to the wave front.

The experiments were performed by sticking a SCBIstretchable grating onto the surface of a cylinder of radiusR = 164 mm machined from an aluminum piece with nooptical finishing of the surfaces. Figure 4a shows a sketchof the optical layout, with the grooves of the grating inthe vertical direction, parallel to the rotation axis of thecylindrical surface. The plane wave front of a collimatedlaser beam impinges onto the surface of the grating, fromwhich it is diffracted. In Fig. 4b a picture of the impinging,reflected and diffracted beams is shown, where the beampaths are evidenced by light scattered from vapors of liquidnitrogen. Due to the cylindrical shape of the surface, thesystem is endowed with a pure astigmatic optical power inthe tangential direction, the curvature being vanishing in theother, so that the sagittal focal length is infinite (see Fig. 4a).The 0th and 1st diffraction orders have been collected witha CCD camera along the corresponding propagation di-rections, and the intensity distributions recorded at severaldistances from the grating. The sizes of the diffracted spotshave been measured and the divergence provided a measureof the tangential focal length of the system. These resultsare compared to that obtained from basic optics: with Rthe cylinder radius, α and β the incidence and diffractionangles, the tangential focal length is given by [31, 43]:

f = Rcos2 β

cos α + cos β(2)

Note that, due to the finite thickness of the gratings,the deformation determines a slight change in the grooves’spacing, from 2.5 ±0.2% to 3.5 ±0.2% depending on thegrating thickness. In Table 1 we report the results of mea-surements obtained from 20 positions for each grating, op-erated at the 0th and the 1st diffraction orders, comparedwith the theoretical expectations.

We have also performed an experiment where the laserbeam is spatially filtered and expanded up to a diameter of


ORIGINALPAPER


Table 1 The effective focal lengths compared to the expected theoretical ones for 5 gratings. E1 and E2 indicate Ag thermallyevaporated gratings (with 123 nm and 60 nm thick metal layers, respectively) while C1, C2 and C3 are the Ag SCBI gratings (with anequivalent thickness of 85, 60 and 140 nm, respectively). These results are the averages of a number of measurements in differentpositions of the gratings, from which we estimated the errors reported here.

Grating 0th-order experimental [mm] 0th-order theoretical [mm] 1st-order experimental [mm] 1st-order theoretical [mm]

E1 9.48 ± 0.73 7.2 8.61 ± 0.89 8.2

E2 8.06 ± 2.48 7.2 7.32 ± 2.53 8.3

C1 7.22 ± 0.47 7.2 8.85 ± 1.11 8.2

C2 7.82 ± 0.90 7.2 8.56 ± 1.46 8.2

C3 7.78 ± 0.62 7.2 8.95 ± 1.57 8.2

Figure 4 (a) The optical layout for the characterization of theoptical power imposed by the cylindrical surface. (b) A picture ofthe optical bench, showing the impinging beam, the reflected andthe 1st diffracted order.

10 mm approximately, then passed through a focusing lensand finally impinging onto the grating well before the focalregion. The focusing conditions are chosen in such a waythat a beam collimated in the horizontal direction emerges

from the grating. The intensity profiles of the first-orderdiffraction beam have been monitored by a CCD camera atdifferent positions along the optical axis, passing throughthe focal plane of the lens. The transversal intensity pro-files showed the expected behavior, the beam being sharplyfocused along the vertical direction only.

4. Conclusions

In summary, we have shown that stretchable reflective op-tical gratings can be easily fabricated by supersonic clusterbeam implantation in PDMS. The reflectance and resilienceof the coating resulting from the implantation of nanopar-ticles in the elastomer guarantee the stability of the opticalperformances upon thousands of deformations of the grat-ing up to the 50% of the original dimensions. Stretchableand conformable optical gratings open novel and very ex-citing perspectives for the fabrication of optical devicesincorporating diffracting and optical power in a single ele-ment. Starting from nonoptical grade curved surfaces, onecan fabricate optical elements by sticking stretchable grat-ings or mirrors fabricated by SCBI metallization of cheapreplica molded and identical PDMS substrates. Further-more, the superior stability and durability of the SCBIstretchable optical elements allows the peeling and reuseof the devices on different surfaces with arbitrary shapeand curvature without compromising their optical quality.

Acknowledgements. The authors would like to thank Marco In-drieri for AFM measurements, Francesco Cavaliere and DanieleVigano for the fabrication of the mechanical supports, AndreaBellacicca for the development of the electronics in the miniatur-ized stretcher and Davide Marchesi for his assistance in the SEMcharacterization. (Supporting Information is available online fromWiley InterScience or from the author).

Supporting information for this article is available free of chargeunder http://dx.doi.org/10.1002/lpor.201300078

Received: 5 June 2013, Revised: 19 July 2013,Accepted: 19 August 2013

Published online: 10 September 2013




Key words: Deformable optics, stretchable optics, nanocom-posite materials, polymeric materials, stretchable gratings.

References

[1] D. R. Williams and J. Porter in Adaptive Optics for Vi-sion Science: Principles, Practices, Design and Applications,edited by J. Porter, H. Queener, J. Lin, K. Thorn, and A. A.S. Awwal, (John Wiley & Sons, Hoboken, NJ, USA 2006),Ch. 1.

[2] B. F. Grewe, F. F. Voigt, M. van ‘t Hoff, and F. Helmchen,Biomed. Opt. Exp. 2(7), 2035 (2011).

[3] O. Albert, L. Sherman, G. Mourou, T. B. Norris, and G.Vdovin, Opt. Lett. 25(1), 52 (2000).

[4] F. Druon, G. Cheriaux, J. Faure, J. Nees, M. Nantel, A.Maksimchuk, G. Mourou, J. C. Chanteloup, and G. Vdovin,Opt. Lett. 23(13), 1043 (1998).

[5] M. Booth, M. Neil, and T. Wilson, J. Opt. Soc. Am. A 19,2112 (2002).

[6] A. Iqbal and F. B. Amara, Int. J. Optomechatron. 1, 180(2007).

[7] E. J. Fernandez, I. Iglesias, and P. Artal, Opt. Lett. 26(10),746 (2001).

[8] H. Oku and M. Ishikawa, Appl. Phys. Lett. 94, 221108(2009).

[9] O. Solgaard, F. S. A. Sandejas, and D. M. Bloom, Opt. Lett.17(9), 688 (1992).

[10] D. W. Carr, J. P. Sullivan, and T. A. Friedmann, Opt. Lett.28(18), 1636 (2003).

[11] M. Aschwanden and A. Stemmer, Opt. Lett. 31(17), 2610(2006).

[12] J. P. Dery, E. F. Borra, and A. M. Ritcey, Chem. Mater.20(20), 6420 (2008).

[13] C. S. Gudeman, C. B. Carlisle, J. Hunter, R. W. Corrigan,R. J. Monteverde, and J. I. Trisnadi, Proc. SPIE 4653, 56(2002).

[14] A. A. Yasseen, S. W. Smith, F. L. Merat, and M. Mehregany,IEEE J. Sel. Top. Quantum Electron. 5, 75 (1999).

[15] C. W. Wong, Y. Jeon, G. Barbastathis, and S.-G. Kim, Appl.Opt. 42, 621 (2003).

[16] R. A. Guerrero, M. W. C. Sze, and J. R. A. Batiller, Appl.Opt. 49(19), 3634 (2010).

[17] J. L. Wilbur, R. J. Jackman, G. M. Whitesides, E. L. Cheung,L. K. Lee, and M. G. Prentiss, Chem. Mater. 8, 1380 (1996).

[18] B. Grzybowsky, D. Qin, R. Haag, and G. M. Whitesides,Sens. Actuators, A 86, 81 (2000).

[19] J. A. Rogers, O. J. A. Schueller, C. Marzolin, and G. M.Whitesides, Appl. Opt. 36(23), 5792 (1997).

[20] Y. Xia, E. Kim, X.-M. Zhao, J. A. Rogers, M.Prentiss, and G. M. Whitesides, Science 273(5273), 347(1996).

[21] I. M. Graz, D. P. J. Cotton, and S. P. Lacour, Appl. Phys.Lett. 94, 071902 (2009).

[22] O. Graudejus, B. Morrison, C. Goletiani, Z. Yu, and S.Wagner, Adv. Funct. Mater. 22(3), 640 (2012).

[23] N. J. Douville, Z. Li, S. Takayama, and M. D. Thouless, SoftMatter 7, 6493 (2011).

[24] C. L. Tuinea-Bobe, P. Lemoine, M. U. Manzoor, M. Tweedie,R. A. D’Sa, C. Gehin, and E. Wallace, J. Micromech.Microeng. 21, 115010 (2011).

[25] N. Bowden, S. Brittain, A. G. Evans, J. W. Hutchinson, andG. M. Whitesides, Nature 393, 146 (1998).

[26] S. Rosset and H. R. Shea, Appl. Phys. A 110, 281 (2013).[27] J. Y. Chung, A. J. Nolte, and C. M. Stafford, Adv. Mater. 23,

349 (2011).[28] G. Corbelli, C. Ghisleri, M. Marelli, P. Milani, and L.

Ravagnan, Adv. Mater. 23(39), 4505 (2011).[29] S. Rosset, M. Niklaus, P. Dubois, and H. R. Shea, Adv. Funct.

Mater. 19, 470 (2009).[30] R. Cardia, C. Melis, and L. Colombo, J. Appl. Phys. 113,

224307 (2013).[31] C. Palmer, E. Loewen, in Diffraction Grating Handbook, 6th

ed. Newport Corporation, USA (2005), Ch. 7.[32] T. K. Gaylord and M. G. Moharam, Proc. IEEE 73(5), 894

(1985).[33] N. Zhu and T. Mei, Opt. Commun., doi: 10.1016/

j.optcom.2013.02.032 (2013).[34] D. Fattal, J. Li, M. Fiorentino, Z. Peng, and R. G. Beausoleil,

Group IV Photonics (GFP), 2010 7th IEEE InternationalConference on, 37 (2010).

[35] D. Fattal, J. Li, M. Fiorentino, Z. Peng, and R. G. Beausoleil,Nature Photon. 4, 466 (2010).

[36] E. Barborini, P. Piseri, and P. Milani, J. Phys. D: Appl. Phys.32, L105 (1999).

[37] K. Wegner, P. Piseri, H. V. Tafreshi, and P. Milani, J. Phys.D: Appl. Phys. 39, R439 (2006).

[38] E. Barborini, P. Piseri, A. Podesta, and P. Milani, Appl. Phys.Lett. 77(7), 1059 (2000).

[39] P. Piseri, A. Podesta, E. Barborini, and P. Milani, Rev. Sci.Instrum. 72, 2261 (2001).

[40] V. Amendola, O. M. Bakr, and F. Stellacci, Plasmonics 5, 85(2010).

[41] K. Chatterjee, S. Banerjee, and D. Chakravorty, Phys. Rev.B 66, 085421 (2002).

[42] M. Aschwanden and A. Stemmer, Proc. SPIE 6524, 65241N(2007).

[43] H. G. Beutler, J. Opt. Soc. Am. 35(5), 311 (1945).


CHAPTER 5

A simple scanning spectrometer based on a stretchablereflective grating

Deformable optical elements (lenses, mirrors and gratings) are fundamental ingredientsfor the fabrication of a novel class of compact, inexpensive and portable devices based onadaptive optics. In particular the use of tunable gratings based on stretchable reflectivesubstrates could revolutionize the design of optical miniature spectrometers and widensignificantly the field of applications. In standard spectrometers a rigid grating spatiallyseparates spectral components of diffracted light which can be either detected simulta-neously by an array of photodetectors, or swept through a photodetector slit by one ormore rotating elements. Both solutions are based on the so-called Czerny-Turner opticalconfiguration [63] allowing a compact design consisting of two concave mirrors and oneflat diffraction grating. This approach is very effective, however it requires the use ofa quite sophisticated and expensive solutions in terms of detectors and/or mechanicalmountings.

In this chapter a low-cost spectrometer operating in the visible range and based onSupersonic Cluster Beam implanted elastomeric diffraction gratings is presented. Themechanical properties of the elastomeric grating are exploited to select the wavelengthof the light diffracted at a given angle respect to the normal of the grating, simply byimposing an increase of the pitch through a stretching of the device. This mount isparticularly demanding in terms of durability of the elastomeric optical element since itneeds to operate at huge number of cycles of strains up to approximately 100% of the ini-tial size while maintaining diffraction efficiency, dispersion performances, and groovesquality. This particular application of elastomeric optical devices can be considered agood benchmark for an extended class of applications, including adaptive optics forspectroscopy, microscopy, and ophthalmology, as seen in the first chapter.

5.1 Grating fabrication

The polymeric substrate for the stretchable grating (30 mm long and 10 mm wide) wasfabricated as described in detail in ref. [23]. Briefly the back of a commercial DVD(grooves pitch 740nm) was used as a master for the molding of the PDMS (Sylgard184 from Dow Corning, in a 1:10 ratio) substrate, obtaining a transparent elastomericdiffraction grating. The transparent PDMS grating was then made reflective by Super-sonic Cluster Beam Implantation of Ag nanoparticles at a rate of 8 · 10−3nm · s−1 upto an equivalent thickness of approximately 60nm. In order to avoid oxidation of thesilver nanoparticles and to facilitate the handling of the device, the metallized gratingwas coated with a second thin layer of PDMS (capping layer) deposited by spin coatingat 3500rpm for 60s, reaching a thickness of about 10µm, and cured at RT for 48 hours.

71

72 5.2 Optical setup

Figure 5.1: Optical setup of the spectrophotometer based on the elastomeric grating

The stretchable reflective grating was then clamped on a homemade stretcher consistingin a computer-controlled stepper motor coupled with a micrometric traslator, the wholesystem able to minimum movements of 2.5µm, as described in [21].

The optical properties of these gratings have been accurately characterized in [23]and they proved to be stable enough for operating after a huge number of strain cy-cles. Performances of these gratings were also compared to those of identical substratescovered by means of traditional evaporation of silver (gold evaporation produces fastdeterioration due to a poor adhesion of gold on PDMS, as shown in [23]).

5.2 Optical setup

The scanning spectrometer based on the stretchable grating is schematically shown inFigure5.1, with the aim of reducing the cost and the number of optical elements in-volved, as well as its dimensions. A light source (LS) has been placed close to a verticalslit (S1) 0.01x5mm2 wide. An achromatic doublet (L1, f = 100mm) produces a beamspatially coherent in the horizontal direction, impinging normally onto the surface ofthe grating (G), with the grooves accurately placed in the vertical direction, after passingthrough the sample (S). The first order diffracted beam, collected by a lens at an angleof 27◦ (L2, f = 100mm), is focused into the collection slit (S2: 0.5x5mm2 wide, verticallypositioned) and then falls onto the detector (D). The detector consists in a cheap, multi-purpose BPX65 photodiode equipped with a custom-made electronics aiming to amplifythe signal while filtering high frequency electrical noises. Such a layout determines anultimate resolution of about 2.5nm overall the wavelength range because of diffractionlimits. The grating is stretched by a stepping motor (M) controlled by a PC. Steps of0.1% up to 100% of the overall grating length have been applied, the stretching beingcontinuously monitored to determine the wavelength of the collected light. The time se-quence of the light intensity is measured synchronously to the stretching, thus obtainingthe sequence of wavelengths sensed by the system.

Note that this system does not need any rotating stage, at variance with traditionalmonochromators, thus requiring a substantially simpler and straightforward mechani-cal mounting. The capability of spanning the whole spectrum by simply stretching thegrating represents the main breakthrough of this system.

A simple scanning spectrometer based on a stretchable reflective grating 73

p1 [nm] p2 [nm] p3 R2

Extension 5.504 · 10−4 −0.1761 −1.4137 0.99986Retraction 5.096 · 10−4 −0.1331 −12.0995 0.99989

Table 5.1: Fit parameters extracted from the quadratic fit of the calibration procedure. R2 repre-sents the regression coefficient.

5.3 Calibration

As a reference for the calibration of the spectrometer we acquired the spectrum of a Hglamp, whose spectrum and emission peak positions are tabulated in [139]. By stretchingthe grating up to 100% all the 5 main emission peaks in the visible range are well de-tectable and can be easily identified. In Figure5.2 we show the results obtained with ourspectrometer operated with the photodiode previously described. The showed spectraare the average of 20 acquisitions, both for extension and retraction of the stretchablegrating. It is worth noting that the extension and retraction cycles do not perfectly co-incide both in wavelength and in the relative intensities. This can be explained withthe relaxation time needed by the polymeric chains of the substrate to rearrange af-ter stretching. As a consequence, different calibrations for the extension and retrac-tion cycles are needed. Data showing the wavelength of the peaks as a function of thestretching applied are reported in Figure5.2b and can be fitted with a quadratic equationε = p1 ·λ2 +p2 ·λ+p3, where λ indicates the wavelength, ε the percentage strain appliedto the grating. The results for the fit coefficient p1, p2 and p3 relative to the extension andretraction cycles are reported in table 5.1 and are used for the determination of the wave-length in the next measurements. The relationship occurring between ε and λ is almostlinear since the quadratic term p1 is very small respect to p2 and p3. Unlike the most ofavailable commercial instruments, our spectrometer can be thus considered as a linearmonochromator. According to this calibration, by stretching the grating up to 100% weare able to span a wavelength range between 350nm and 620nm with minimum steps of0.3nm (i.e. 0.1% of strain).

The spectral resolution is approximately 4 − 5nm, close to the ultimate resolution ofthe optical device. The resolution is limited by the non-perfect shape of the reflectingsurface that is affected by small local deformations changing with the stretching force.Works are in progress to understand the limitations imposed by these features that areultimately related to the flexible nature of the elastomeric substrate. Moreover, since thegrating consists in the replica of a DVD surface, some defects on the grooves structurecan be present and the grooves are not linear but curved. The light spot impinging onthe grating has a diameter of 6 millimeters and thus the diffracted beams, entering thephotodiode through the vertical linear slit, are affected by the curvature of the gratinglines, limiting the final resolution of the spectrum.

5.4 Spectra acquisition

As a benchmark, we have performed the analysis of the extinction spectrum of Rho-damine B dye, presenting a sharp and highly intense absorption peak at 543.6nm [140].The extinction spectrum has been acquired and compared with the spectrum of the samesample acquired with a commercial spectrophotometer (Jasco 7850) and with data re-ported in literature [140]. A 3.2 · 10−4M solution of Rhodamine B in ethanol has beenprepared in a quartz cuvette with an optical path of 10mm for the analyses.

74 5.4 Spectra acquisition

Figure 5.2: (a) Hg lamp spectra taken with the SCBI grating stretched up to 100%. Extension andretraction spectra are not coincident due to delay in the response of the polymer when released.Square and circles highlight the peaks, for extension and retraction spectra respectively, used forthe calibration of the device while numbers indicate the reference wavelengths from [139] (b) Peaksposition in terms of grating stretching (black circles for extension, red squares for retraction) andquadratic fit (solid black line for extension, red dotted line for retraction).

Extinction spectra of Rhodamine B have been acquired by stretching the grating upto 100% and acquiring first the light intensity transmitted by the ethanol (I0(ω), the ref-erence spectrum), then the light intensity transmitted by the dye (I(ω)). The extinctionspectra A(ω) have been calculated as follows:

A(ω) = − log10

I (ω)

I0 (ω)(5.1)

The intensities I0(ω) and I(ω) used in equation 5.1 for the determination of the ab-sorbance are the average of 20 spectra both for ethanol and rhodamine B solution. Thebaseline values of the two spectra were measured as the mean of 1000 intensity valuesacquired applying a 100% strain to the grating, and subtracted from I0(ω) and I(ω). Theobtained absorbance spectra were then appropriately smoothed by means of a splineprocedure.

The results in Figure5.3a show peaks at 545.0nm and 545.3nm for the extension andretraction cycles respectively, in good agreement with the data reported in literature([140], black continuous line). Despite the differences in the spectra of the Hg lampshown in Figure5.2, the two spectra of the Rhodamine B acquired during the exten-sion and retraction cycles coincides, thanks to the different calibrations used in the two

A simple scanning spectrometer based on a stretchable reflective grating 75

Figure 5.3: (a) Absorption spectra for Rhodamine B dye with a concentration of 3.2 · 10−4M . Ex-tension and retraction spectra acquired with the photodiode are perfectly coincident (solid greenand blue lines) and the peak wavelength of 545.0nm and 545.3nm are comparable with the wave-length peak of reference spectrum [140] (dotted black line, 543.6nm). The same sample was alsoanalyzed with a commercial Jasco 7850 Uvvis spectrophotometer (dash-dot red line), showing apeak position of 548.0nm. The broadening of the spectra taken with SCBI grating is mainly due tothe curvature and slight tilt of the diffracted lines caused by the geometry of the grating groovesthat is replicated from a DVD with curved lines. (b) The same extension and retraction absorptionspectra with the associated uncertainties (dashed red and blue lines).

cases. The spectrum acquired with the commercial spectrophotometer is characterizedby a shape similar to the one reported in literature, with a shift in the peak positionto 548.0nm. The slightly larger broadening of the spectra obtained with the stretchablegrating can be explained with the nature of the grating itself, as mentioned above.

Similar absorbance peak intensities are measured between the extension and retrac-tion cycles and the spectrum measured with the spectrophotometer as well. The smalldifferences of the peak intensity abundantly fall in the uncertainty, showed in Figure5.3b.Here only the extension and retraction spectra with the relative uncertainties are shown,with the same scale of Figure5.3a, for sake of clarity. The uncertainties are obtained bypropagating the point-by-point standard deviation of the ethanol and rhodamine spec-tra in the extension or retraction cases. It is worth noting that the uncertainties largelyincrease for shorter wavelengths (and thus for low stretching percentages). This behav-ior is due to the measure of the baseline subtracted from the original averaged ethanoland rhodamine spectra and to the high electrical noise detected by the photodiode dur-ing spectra acquisition. In Figure5.4a the averaged ethanol and rhodamine extensionspectra before the baseline subtraction are reported. At low stretching percentages thetwo spectra should coincide since the transmittance of rhodamine is expected to be uni-tary. However a slight difference is observed, as can be clearly seen in the magnificationof the first part of the spectra shown in Figure5.4b. This difference, of the order of ap-proximately 2 · 10−4V , can be eliminated by subtracting the baselines relative to the twosamples. Nevertheless the electrical noise read by the photodiode during the acquisi-tion of the baseline, shown in Figure5.4c, is high, being the fluctuations of intensity evenlarger than the difference between the two spectra, of the order of 4 · 10−4V . This maylead to larger uncertainties of the first part of the absorption spectrum, once the mean

76 5.4 Spectra acquisition

Figure 5.4: (a) Averaged extension and retraction spectra for ethanol (blue line) and rhodamine(red line) before the baseline subtraction, and a magnification of the first stages (b) highlightingthe difference between the spectra. (c) The rhodamine baseline characterized by fluctuations ofthe same order of magnitude of the difference of spectra.

value of the baseline is subtracted. The uncertainty on the absorption peak region ismuch lower due to the larger difference between the ethanol and rhodamine spectra, atleast one order of magnitude larger than the baseline fluctuations (signal-to-noise ratioof approximately 10). The same considerations can be applied to the retraction spectra.

In conclusion we have built an extremely cheap and simple spectrometer based uponthe novel technology of SCBI stretchable gratings, capable of spanning the entire visi-ble range of wavelengths without any moving component except the grating stretcher,and maintaining its optical performances for thousands of strain cycles. These resultssuggest the possibility to develop a novel class of cheap spectrometers, proper for allthe applications where the resolution is not extremely demanding. The analysis of theobtained results suggests the directions toward the improvement of the system. The op-timization of the optical setup, the molding of gratings characterized by straight groovesand their better clamping on the stretcher can lead to an improvement of the attainablespectra resolution. The uncertainties can be minimized, and thus the intensity preci-sion enhanced, simply by acting on the noise filter and signal amplification stages of theelectronics controlling the photodiode.

Part III

Plasmonics

CHAPTER 6

Surface Plasmon Resonance (SPR)

In the previous chapters peculiar behaviors of the optical properties were observed, inparticular regarding the non-monotone trend of reflectivity with the amount of nanopar-ticles supersonically implanted in the Ag/PDMS nanocomposite used for the fabricationof elastomeric mirrors or diffraction gratings. The interaction between the metal and theoscillating external electromagnetic field affects the optical response of metallic mate-rials [141]. The effect of this interaction is enhanced for metal particles with nanomet-ric size [100], as in the case of metal-polymer nanocomposites used for the fabricationof elastomeric optical devices, and can be responsible for the behavior observed in re-flectance spectra. Electromagnetic radiation impinging on the nanocomposite excitesthe free electrons of the metal, that begin to oscillate [142]. If the frequency of the inci-dent light matches the characteristic frequency of the electrons-positive core interaction,a propagating resonant effect called plasmonic resonance arises [143].

The surface-to-volume ratio of small metallic nanoparticles is much larger than of bulkmaterials and the electronic states modify accordingly, with the arising of electronic sur-face states. As a result the free electrons are excited at different (usually higher) frequen-cies. This phenomenon is called Surface Plasmon Resonance (SPR) [100] or Localized SurfacePlasmon Resonance (LSPR) if the size of the nanoparticle is much lower than the wave-length of the plasmonic wave, thus remaining localized on the particle, as depicted inFigure6.1. Surface Plasmon Resonance phenomena can be clearly measured by meansof light extinction spectra in the visible range since resonances typically involve a largeabsorption of energy and thus light intensity.

Plasmons provide the basis for color-changing biosensors [144, 145], photothermalcancer treatments [146, 147], improved photovoltaic cells efficiency [148, 149, 150] andnanoscale photonics circuitry [151, 152]. In these applicative fields the strong enhance-ment of the electric field caused by the excitation of the plasmon resonance is exploited.

Figure 6.1: Picture of the free electrons-light interaction casing the surface plasmon resonance formetal nanoparticles.

79

80 6.1 Theoretical models

Surface-enhanced Raman spectroscopy (SERS) is a surface-sensitive technique enhancingRaman scattering by molecules adsorbed on a surface properly functionalized in orderto present a plasmon resonance [153]. When the incident light in the experiment strikesthe surface, localized plasmon is excited and the strong electric field generated in turnexcites the molecules adsorbed on the surface. If the shift is small, the Raman signalof the molecules excites again the plasmon resonance and a strong signal is emitted bythe surface. Thus the signal is amplified by a factor E4, allowing the detection of singlemolecules [154]. This is particularly convenient in the biosensing applicative field [144,155, 156]. Gas sensing represents another interesting application of surface plasmonresonance [157, 158, 159, 160].

The strong electric field generated by the resonance can also be exploited in photoniccircuits [151] for the amplification of the emission of a gaining medium surrounding theexcited metal nanoparticle. Such systems represent a surface plasmon amplification by stim-ulated emission of radiation (SPASER), that is the smallest laser source to date [161, 162].Nano-light sources in plasmonic circuits can be coupled with plasmonic waveguides andphotodiodes or nanoantennas [163, 164, 165].

In order to finely tune the properties of functional plasmonic substrates for the citedapplications and to extract information on the metal-polymer nanocomposite system,used in this thesis work, from optical characterization, a physical modeling of the reso-nances is necessary. In this chapter an overview of the main models describing surfaceplasmon resonance is given.

6.1 Theoretical models

Surface plasmon resonance causes the appearance of strong peaks in the light absorptionspectrum A(ω) of the sample. Light absorption in a medium is related to the absorptioncoefficient α(ω) and the thickness ` of the medium by means of the Lambert-Beer law:

A(ω) = α(ω)` = Nσ(ω)` (6.1)

Where N is the number of absorber (in our case the number of particles) and σ(ω) theextinction cross section, accounting for the plasmon resonances. The physical parametercontrolling the behavior of the extinction cross section is the global dielectric function εof the absorbing mean [100], that must therefore be calculated in order to foresee orunderstand the optical response of the material.

6.1.1 Drude model

Considering the free electrons like a classic gas moving freely and interacting with abackground of positive ions much heavier than electrons, and thus considered fixed,represents the simplest approach for the calculation of the dielectric function of a metallicmaterial. Three main assumption are introduced:

1. Free electrons approximation: The interaction between free electrons and the pos-itive ions background is negligible. The only role of positive ions consists in main-taining the overall electrical neutrality of the material.

2. Independent electrons approximation: Interaction between electrons is neglected.

3. Relaxation time approximation: Collisions between electrons do not depend onthe result of previous collisions.

Surface Plasmon Resonance (SPR) 81

These strong assumptions are however satisfied by materials with complete internalelectronic states and with a single electron in the outer shell contributing to conduc-tion (that can be considered free). Alkali metals (Na, K, Rb, Cs) perfectly represent thisclass of materials. This theory can also be exploited for noble metals, introducing propercorrections accounting for intraband transitions.

The behavior of free electrons subjected to an external electrical field ~E(t) is describedby the following equation of motion [143]:

med2~x(t)

dt2+meγ

d~x(t)

dt+ e ~E(t) = 0 (6.2)

whereme is the electrons mass and γ = 1τ represents the damping parameter, accounting

for electrons collisions, and proportional to the inverse of the electron relaxation timeτ . The latter is the time between two subsequent collisions involving the consideredelectrons. The electric field ~E(t) of an electromagnetic radiation oscillates at a givenfrequency ω and can therefore be written as ~E(t) = ~E0 · exp(−iωt). If the mean free pathl of the electrons is much smaller than the wavelength λ = 2πc

ω of the electromagneticradiation (l � λ), the electric field felt by electrons can be considered constant and thequasi-static approximation condition is satisfied.

If an oscillating electric field is considered, a particular solution of the equation ofmotion is given by ~x(t) = ~x0(t) · exp (−iωt). Substituting this particular solution inequation 6.2, one obtains:

~x(t) =e

me (ω2 + iγω)~E(t) (6.3)

x0 =e

me (ω2 + iγω)E0 (6.4)

From classic electrostatics, the electrical displacement ~D, the electric field ~E and themacroscopic polarization ~P = −ne~x induced by the collective motion of the electroniccloud, are bonded by the following relationship:

~D = ε0 ~E + ~P = ε0ε ~D (6.5)

Gathering equations 6.5 and 6.3, the following result can be easily obtained:

~D = ε0

(1−

ω2p

ω2 + iγω

)~E = ε0ε(ω) ~E (6.6)

where ωp is called plasma frequency, the frequency at which the electromagnetic field isresonant with the electrons oscillation and the absorption peak in spectra arises:

ωp =

√ne2

ε0me(6.7)

with n the free electrons density per unit volume. This plasma frequency depends onthe density of free electrons in the conduction band and is therefore different for eachmaterial. The quantity in brackets in 6.6 coincides with the complex dielectric function ofthe material:

ε(ω) = 1−ω2p

ω2 + iγω=

(1−

ω2p

ω2 + γ2

)+ i

(ω2pγ

ω (ω2 + γ2)

)(6.8)


Figure 6.2: The real and imaginary parts of ε(ω) for silver determined by Johnson and Christy[167] (red dots) and a Drude model fit to the data [166].

This result well describes the behavior of metals in which electrons can be consideredas ideally free. The behavior of noble metals (Ag among them) differ substantially fromthe theoretical expectations. These differences can be ascribed mainly to two effects: thecontribution of the positive ions background to polarization and the interband transi-tions occurring at high frequencies. Moreover, corrections must be introduced in orderto consider the finite size of metal particles.

Core polarization

Free electron approximation is no longer valid for noble metals since the partially filled dband close to the Fermi energy surface causes a highly polarized environment, affectingthe motion of the electrons in the conduction band [166]. From Figure6.2 a significativedifference of the experimental dielectric function ε(ω) respect to the expected theoreticalvalues calculated with the Drude model at high frequencies (higher than 3eV) is clear.The effect of the polarized environment caused by the partially filled d band can beaccounted by adding a term εcore to the macroscopic polarization vector:

~Pcore = ε0 (εcore − 1) ~E (6.9)

thus obtaining the following dielectric function:

ε(ω) = εcore −ω2p

ω2 + iγω(6.10)

εcore is usually determined by comparing experimental data of the real and imaginaryparts of the dielectric function ε(ω) with theoretical models, and it takes values in therange 1 ≤ εcore ≤ 12.

Interband transitions

ε(ω) in free electrons metals is governed mainly by electronic transitions within the con-duction band or interband transitions from inner to conduction band or between the con-duction band and higher unoccupied states. In particular noble metals represent mono-valent metals in which both the types of transition take place and depend on the elec-tronic structure of noble metals. Metallic materials like, for example, Cu, Ag, or Au, havecompletely filled 3d, 4d and 5d shells and just one electron in the 4s, 5s and 6s bands re-spectively. The interband transitions occur within the broad conduction band (derivedmainly from Ag 5s1p hybridized atomic orbitals), which onset at zero frequency [168].


Interband transitions in noble metals can be accounted by introducing an additionalcomplex term to susceptibility χIB = χIB1 + iχIB2 (and as a consequence to the dielectricfunction, since χ = ε − 1). This term can be calculated by considering the transitionmatrix element 〈f |M |i〉 between the initial and final states of the electron. From thesemiclassical model of Parravicini [169], the imaginary part of complex dielectric sus-ceptibility of the interband transitions satisfies the following relation:

ω2χIB2 (ω) ∝ 2

(2π)3

∑i,f

|〈f |e~r|i〉|2∫BZ

d3~kδ[Ef

(~k)− Ei

(~k)− ~ω

](6.11)

〈f |e~r| i〉 is the matrix element of the dipole operator between the initial and final states ofthe electron, with energiesEi andEf respectively. The integration is done with respect tothe components of the wave vector ~k in the Brillouin zone. The real part of the dielectricsusceptibility for interband transitions can be obtained by using the Kramers-Kronigtransformations. The effects of the interband transitions on the dielectric functions canbe accounted in the εbulk term, including the previously introduced for the effect of thepositive ions background, and obtainable by the comparison between the theoreticalexpectations and experimental values [170].

Finite size effect

The model seen in this section is relative to an infinitely extended metallic material, orwith dimensions much larger than the free mean path of the free electrons, of the order tofew tens of nanometers. This assumption is no longer valid if nanometric-size particlesare considered, as in the case of the silver nanoparticles used in this work.

The relaxation damping term γ appearing in the fundamental equation 6.8 concernsthe limitations of the electron mean free path due to collisions between the electrons andof electrons with phonons o impurities of the material. In general the damping effect canbe written as:

γ → γ∞ =vFl∞

(6.12)

where vF is the Fermi velocity and l∞ the mean free path of electrons in the bulk metal(approximately 52nm in the case of bulk silver [171]). If we have small particles, witha size R � l∞ the electrons mean free path is limited by the dimension of the particles[172, 173] and an effective mean free path lR < l∞ must be introduced. In general lR isproportional to the nanoparticle size R [100, 166, 168] and thus we can write:

γ(R) = γ∞ + ∆γ(R) with ∆γ(R) = AvFR

(6.13)

Here A is a proportionality coefficient accounting for the aspect ratio, the electrons scat-tering isotropicity (for spherical particles is unitary) [168] and scattering from the par-ticle surface [174]. This term can be added to the dielectric function 6.8, obtaining thefollowing result:

ε(ω,R) = εbulk(ω) + ω2p

(1

ω2 + γ2∞− 1

ω2 + γ(R)2

)+ i

ω2p

ω

(γ(R)

ω2 + γ(R)2− γ2∞ω2 + γ2∞

)(6.14)

εbulk(ω), whose values can be found in [167, 175], considers both the interband transi-tions and the core polarization effect


6.1.2 Mie theory

The Drude model can be exploited in the framework of the Mie electrodynamical the-ory describing the optical response of non-interacting metal spheres irradiated by anexternal electromagnetic field [176]. Maxwell equations are solved with proper bound-ary conditions in spherical coordinates and exploiting the multipoles expansion of thepotential generating the external field [100]. Another assumption of Mie’s theory con-sists in considering both the particles and the surrounding medium as homogeneousand describable by their bulk optical dielectric functions [177, 178].

Since we deal with nanoparticles with a size of some nanometers (see section 3.2.1 fora silver nanoparticles size distribution characterization), the condition R << λ is satis-fied and the quasi-static approximation can be applied. Only the first term of the multipoleexpansion, due to dipole excitation, is considered. Out of the approximation higherorders of the multipole expansion must be considered, introducing retardation effectsaffecting the optical response of the system, in particular the plasmonic absorption peak.

Radiation-matter interaction can be modeled considering two main mechanisms:light absorption with generation of heat, and light scattering. The sum of absorption andscattering is equivalent to extinction. The two different interaction mechanisms are de-scribed in terms of the interaction cross-sections, related to the radiation intensity loss∆I(z) through the Lambert-Beer law:

∆Iabs(z) = I0(1− e−ncσabsz

)(6.15)

∆Isca(z) = I0(1− e−ncσscaz

)(6.16)

σext = σabs + σsca (6.17)

where nc is the nanoparticles density and σabs, σsca and σext the absorption, scatteringand extinction cross-sections respectively. In the framework of the Mie’s theory, thecross-sections can be expressed as follows [166]:

σext =2

x2

∞∑L=1

(2L+ 1) [<(aL + bL)] (6.18)

σsca =2

x2

∞∑L=1

(2L+ 1)[<(|aL|2 + |bL|2

)](6.19)

x depends on the particle size R and on the refractive index of the surrounding mediumnm:

x =2πRnmω

(6.20)

aL and bL are the scattering coefficients and can be expressed in terms of the Ricatti-Bessel functions ψL(x) and ξL(x):

aL =mψL(mx)ψL′(x)− ψL(x)ψL′(mx)

mψL(mx)ξL′(x)− ξL(x)ψL′(mx)(6.21)

bL =ψL(mx)ψL′(x)−mψL(x)ψL′(mx)

ψL(mx)ξL′(x)−mξL(x)ψL′(mx)(6.22)


and m = nnm

the ratio between the refractive indexes of the particle and of surroundingmedium. The subscript L indicates the order of the spherical multipole excitation: L = 1refers to dipole oscillation, L = 2 to quadrupole oscillation and so on. In quasi-staticapproximation only the first term (i.e. the dipole approximation) is considered and thusit follows [166]:

σsca =8π

3

∣∣∣~k∣∣∣R6

∣∣∣∣ ε− εmε+ 2εm

∣∣∣∣2 σabs = 4π∣∣∣~k∣∣∣R3=

(ε− εmε+ 2εm

)(6.23)

Where ε and εm are the dielectric functions of the nanoparticle and of the surroundingmedium respetively, and |~k| = 2π

√εmλ the wave vector of the incident light. From the

different dependence of the two cross-sections from the particle size R it is clear that forextremely small particles (i.e. nanoparticles) the scattering cross-section can be neglectedand thus the only contribution to extinction is due to absorption, σext ' σabs. Hence forsmall particles in dipole approximation and considering the complex dielectric functionε = ε1 + iε2:

σext = σabs = 9ω

cε

32V0

ε2

[ε1 + 2εm]2

+ ε22(6.24)

where V0 = 43πR

3 is the volume of the particle. From equation 6.24, the extinction cross-section is characterized by a resonance in the case ε1(ω) = −2εm and does not directlydepend on the cluster size R but only indirectly through the size dependance of thedielectric function ε(ω,R), as discussed in the previous section and shown in equation6.14.

If ω � γ, the damping term can be neglected, and substituting equation 6.8 in 6.24, alorentzian curve, typical of absorption peaks, is obtained:

σabs = σ01

(ω − ωMie)2

+(γ2

)2 (6.25)

where ωMie is the Mie’s resonance frequency than, in dipole approximation, can be ex-pressed as:

ωMie =ωp√

1 + 2εm(6.26)

If the polarization of the positive ion background is considered [100]:

ωMie =ωp√

εcore + 2εm(6.27)

Mie’s theory well describes the optical behavior of non-interacting metal nanoparti-cles but is an electrodynamical model, giving no insight in the physics of the system. Forexample, it does not take into account the effects due to the matrix and the polydisper-sion of the particles size [168]. More important, in metal-polymer nanocomposites syn-thesized by SCBI, metal nanoparticles are not well isolated as can be seen in Figure3.9,but their concentration depends on the implanted dose [22]. Hence metal nanoparticlesfeel the electrical field generated by the other particles and the non-interacting particlesassumption is no longer valid. Nanoparticles implanted in the polymer in high con-centrations aggregate, leading to the formation of metal nanostructures with differenttopologies and not uniformly distributed inside the dielectric. As a result, a differentmodel, considering the macroscopic behavior of the nanocomposite system as an effec-tive medium instead of single particles, and giving an insight in the physics of the system,is desirable.


6.1.3 Maxwell-Garnett theory

In nanocomposite systems with a high concentration (or volume fraction, defined by equa-tion 3.4) of nanoparticles, their reciprocal distance is such that reciprocal interactions cannot be neglected anymore [179, 168]. In particular if the distance between the nanopar-ticles drops below 5R, a redshift and a broadening of the absorption peaks is observedand enhanced by the formation of cluster aggregates. As an example, an increase of theresonance peak position from 520nm up to 750nm is observed if two gold particles areapproached down to a distance of 0.5nm [180]. The Mie’s electrodynamical theory failsto describe such a behavior.

Maxwell Garnett theory [181] is an effective medium theory describing the macro-scopic behavior of a system from the knowledge of the physical properties of its con-stituents. This theory is valid only in the quasi static approximation (2R � λ) and forvery small interparticles distance. In this context a statistical description of the system,taking into account the distribution of the particles in the dielectric, their reciprocal dis-tance and their aggregation status, is preferable, and the volume fraction f representsthe parameter discriminating the different configurations of the system. For f � 1 clus-ters are so diluted that the interactions between them can be neglected: in this limit thetheory coincides with the Mie’s model and the optical response of the whole materialis given by the sum of the contributions of each particle treated singularly. For largervalues of f it makes no sense to deal with single non interacting particles and neither thedielectric function εm is no longer valid to describe the dielectric medium between theclusters.

In the framework of the Maxwell Garnett theory, the dielectric medium between theparticles is conceptually replaced with an effective medium, with a dielectric function εeff ,accounting for all the electrostatic interactions between neighboring clusters. εeff canbe inferred from the dielectric functions ε characterizing the metal particles and εm ofthe dielectric matrix, and the volume fraction f . Extinction of the system is due to theeffective electromagnetic field EMaxwell acting on a given particle and by the individ-ual particle properties (i.e. its polarizability α), but if the concentration of the metallicfiller is sufficiently high the field felt at a given particle position is affected by the fieldsgenerated by the other particles and by the polarizability of the medium until a certaindistance (see Figure6.3). Given a reference particle, the sphere containing all the parti-cles contributing to the optical response of the former is called Lorentz sphere. Hence at agiven position the local electric field is:

Elocal = EMaxwell + Es + Enear (6.28)where Es is the field generated by the polarization charges at the Lorentz sphere andEnear the electric field generated by neighboring particles. Under quasi-static assump-tion, Es can be calculated and is equal to:

Es =P

3ε0εm(6.29)

with P representing the macroscopic polarizability of the dielectric:

P = Elocal∑j=1

Nnjαj = ε0 (εeffεm)EMaxwell (6.30)

nj is the number density of the j-th particle with polarizability:

αj =ε− εmε+ 2εm

R3j (6.31)


Figure 6.3: The Lorentz sphere concept applied to cluster matter. The signs of the charges corre-spond to an empty sphere in the dielectric surrounding [100].

Thus, in the quasi-static approximation, the whole system can be considered homoge-neous and as an effective medium with a dielectric function εeff = ε1,eff + iε2,eff that,combining these last results, is given by:

εeff (ω) = εm1 + 2

3ε0εm

∑j njαj

1− 13ε0εm

∑j njαj

(6.32)

This effective dielectric function can be rewritten in terms of physical measurable quan-tities, like the volume fraction, obtaining the Maxwell Garnett formula:

εeff = εm1 + 2fΛ

1− fΛ, Λ =

1

εm

ε− εmε+ 2εm

(6.33)

This effective dielectric function reaches a maximum value (i.e. a resonance) if the fol-lowing condition is satisfied [182]:

ε(ωMGspr

)(1− f) + εm (2 + f) = 0 (6.34)

Exploiting the equation 6.8 for the dielectric function in the Drude model, the resonantwavelength of the Maxwell Garnett theory can be obtained:

ωMGspr (f) =

ωp√(2+f1−f

)εm + εcore

(6.35)

that, for f → 0 coincides with the result obtained in the Mie’s theory, while it is clear thatthe resonance wavelength strongly depend on the filler volume fraction.

In order to better understand and justify the experimental results that will be pre-sented in the next chapter, a study of the effect of the different parameters on the opticalresponse of the system according to this model is necessary. The nanocomposite synthe-sized by SCBI is characterized by nanoparticles with a given distribution size, as shownin Figure3.8 and thus with different optical responses when irradiated with visible light.


In general absorbance Atot is the sum of all the contributions due to different clustersizes:

Atot = N∑i

ni (Ri) kλi (Ri) t (6.36)

where N is the number of nanoparticles in the irradiated nanocomposite portion, ni(Ri)the fraction of nanoparticles with size Ri and absorption coefficient kλi(Ri), and t thenanocomposite thickness (i.e. the nanoparticles penetration depth). The absorption co-efficient can be easily calculated from equation 6.24 [100] and expressed in terms of thewavelength λ for a direct comparison with the measurements:

kλ = 18πfε32

=εeff (λ)

λ(

(<εeff (λ) + 2εm)2

+ =εeff (λ)2) (6.37)

where the dielectric function in this case is the effective dielectric function ε(λ) = εeff (λ)obtainable from equation 6.33. In the effective dielectric function, the dielectric functionof the metal nanoparticle (in this case silver) is obtained by equation 6.8 re-expressed interms of the light wavelength:

ε(λ) = εbulk(λ)− 2πcτλ2

λ2p (2πcτ + iλ)(6.38)

It is worth noting that the absorption coefficient kλ in equation 6.37 does not depend onthe particle size, thus, according to the Maxwell Garnett theory, R does not affect neitherthe position nor the shape of the plasmonic absorption peak in an absorption spectrum.

The parameters needed for the plot of the theoretical spectra are obtained from lit-erature or from direct measurements part of this work. A silver nanoparticle - PDMSnanocomposite is considered, simulating the system studied in this work and shown insection in the TEM image in Figure3.9. PDMS is a highly transparent elastomer and thusthe imaginary part of the dielectric constant εm is negligible. From the PDMS datasheet[183], the (real) dielectric constant is εm = 2.7. The plasma resonance peak for bulk silveris calculated with equation 6.7 and turns out to be λp = 136nm [184], and the relaxationtime τ = 8.6fs obtained from [185]. The bulk silver dielectric function was extractedfrom an interpolation of data taken from Palik [175] in the range between 350nm and900nm. The real and imaginary parts of εbulk are reported in Figure6.4.

In Figure6.5 the simulated curves of the plasmon peak according to Maxwell Garnetttheory are reported for different values of volume fraction f . The curves are obtainedsimply by using the absorption coefficient 6.37, since it does not depend on the clus-ter size. By increasing the volume fraction a redshift of the resonance frequency is ob-served, together with a sharpening and intensity increase of the peak. The trend of theresonance frequency as a function of the volume fraction is reported in the inset of Fig-ure6.5: data from [175] were fitted with a quadratic polynomial, obtaining the equationλspr = 277.76f2 − 45.63f + 397.73. The discontinuities of data in this graph are due tothe interpolation procedure of the bulk dielectric function.

6.1.4 Shell-core model

The models seen in the previous sections describe the optical behavior of homomolecularparticles, i.e. the behavior of particles constituted by just one element (in this case, sil-ver). However it is well known that silver is subject to oxidation [186] even if implantedin an elastomeric matrix, since PDMS is permeable to air [187]. In ambient atmosphere


Figure 6.4: Plot of the interpolated data (taken from [175] for the bulk dielectric function εbulkrelative to silver. Black circles represents experimental data, interpolated with the red line.

Figure 6.5: Simulation results of the plasmon peak calculated according to Maxwell Garnett theory(equation 6.37) for different values of the volume fraction f . The inset shows the trend of theresonance frequency as a function of the volume fraction (black circles), fitted with a second-orderpolynomial (red line).


silver particles oxidize on the surface only [188, 189] according to the following reaction[186]:

4Ag +O2 → 2Ag2O (6.39)

and oxide layer creates a passivation film protecting the bulk against oxidation. This re-action gives rise to the formation of eteromolecular clusters with a pure silver core and asilver oxide shell growing on the surface with a lower density (7.14g/cm3) compared topure silver (10.49g/cm3) and thus increasing the total size of the nanoparticle. For suchsystems classical Mie’s or Maxwell Garnett theories are not sufficient and a shell-coremodel must be used [190]. Silver nanoparticles oxidation affects the cluster polarizabil-ity α: if Rs, Rc represent the radius of the particle (core + shell) and of the silver corerespectively, with dielectric functions εs and εc, then it follows from [190]:

α = 3V0εm

(RsRc

)3(2εs + εc) (εs − εm)− (εs − εc) (2εs + εm)(

RsRc

)3(2εs + εc) (εs + 2εm)− (εs − εc) (εs − εm)

(6.40)

This polarizability must be used in the following absorption cross-section in order todescribe the plasmon resonance peak of the shell-core particle:

σabs =2π

λ√εm=α (6.41)

This model was exploited in [191] for the modeling of the behavior of silver nanopar-ticles dispersed in polystyrene and oxidized with thermal annealing treatments. In [191]the oxidation of the silver nanoparticles gives rise to a two-peaks structure in the ab-sorption spectrum: a first peak at a wavelength around 400nm due to pure silver par-ticles, and a second contribution at higher wavelengths, convoluted with the first, dueto the oxidized particles and describable with the shell-core model. The contributionof oxidized silver particles redshifts with the increase of the annealing temperature andthe consequent thicker oxide shell: this behavior was studied extensively in [192] byFTDT simulations. The redshift of the oxidized particles peak is explained in terms ofthe decrease of interparticle distance caused by the increase of the oxide shell, also in-creasing the total volume of the cluster because of the lower density respect to puresilver, as shown in Figure6.6a. By exploiting equations 6.41 and 6.40, absorption spectrawere simulated for different RsRc ratios in order to better explain the results shown in thenext chapter regarding Ag/PDMS nanocomposites synthesized by SCBI. The same in-put parameters used in the case of Maxwell Garnett theory were employed in this casetoo. Additionally, the dielectric function of silver oxide εs was taken and interpolatedfrom [175] in the range between 350nm and 900nm (Figure6.6b), and the cluster radiusobtained from the TEM characterization of the size distribution reported in Figure3.8.The results of the simulations, presented in Figure6.6c, show a redshift and intensitydecrease of the plasmonic peak with the growth of the silver oxide shell around theparticle, compatible with a decrease of the interparticle distance observed, for example,in [182, 180, 193]. The inset of Figure6.6c shows the results for the trend of the plas-monic peak wavelength as a function of the Rs

Rcratio, fitted with the quadratic polyno-

mial λspr = −91.97(RsRc

)2+ 355.37

(RsRc

)+ 218.15.


Figure 6.6: (a) Shell-core model with decreasing interparticle distance due to the growth of thesilver oxide on the particle surface [192]. (b) Real and imaginary parts of the dielectric functionof Silver oxide Ag2O (black circles) interpolated with the red line. (c) Simulation results of theplasmon peak calculated according to the shell-core model (equation 6.41) for different values ofthe Rs

Rcratio. The inset shows the trend of the resonance frequency as a function of the particle/core

radius ratio (black circles), fitted with a second-order polynomial (red line).

6.2 Bergman model

In the previous sections we have seen that a dielectric matrix filled with metal nanopar-ticles at a concentration such that the reciprocal interactions can not be neglected, canbe described by introducing an effective dielectric function εeff . The dielectric functiondescribing the behavior of a matrix filled with metal nanoparticles is obtained startingfrom the dielectric functions of the constituents of the system, properly combined bymixing formulas like equation 6.33 for Maxwell Garnett theory. According to this theoryonly systems characterized by low concentrations can be considered, such that the recip-rocal interactions between the nanoparticles can be neglected. Other theories have beendeveloped for the description of the optical behavior of metal-polymer nanocomposites.In particular the Bruggeman theory overcomes the limits of the Maxwell Garnett the-ory in terms of validity for volume fractions f > 0.3 and of the nanoparticles shape,that is assumed to be spherical. However also the Bruggeman theory suffers of somelimitations: some resonances for ε1

ε2< 0 (with ε1 and ε2 the dielectric functions of the

constituents of the nanocomposite) are not predicted and, more important, the effectivedielectric function εeff can be calculated exactly only for very simple arrangements (ortopologies) of the two phases inside the composite material [194].

The Bergman representation [194] represents a more general approach for the effec-tive description of two phases composite systems, independently on the nanoparticlessize, for arbitrary volume fraction and for each possible system topology or combination

92 6.2 Bergman model

Figure 6.7: Schematic drawing of a parallel plate condenser of thickness L, filled with a compositemedium made of arbitrarily shaped homogenous grains with a dielectric function ε2 embeddedin a homogeneous host with a dielectric function ε1 [194]. ~E is the electric field across the samplegenerated by the difference of potential φ(0)− φ(L).

of topologies. As in the case of the Mie’s theory, Maxwell’s equations are applied, inthe quasi-static approximation, to a system consisting in a metallic filler with dielectricfunction ε2 and volume fraction f inside a dielectric matrix with dielectric function ε1,sandwiched between two electrodes through which an electric field ~E(ω) is applied (likea capacitor, see Figure6.7). Thus the effective dielectric function εeff is defined throughthe expression for the electrostatic energy stored in the composite material:

1

2ε0εeff (ω)

⟨~E(ω)

⟩2=

1

V

∫V

1

2ε0ε(~r, ω) ~E2(~r, ω)d~r (6.42)

where ε(~r, ω) represent the dielectric function of the constituent in a given position ~rof the system. This integral can be developed by integrating on the whole volume ofthe composite material, applying the proper boundary conditions. All the geometricalconsiderations and topology of the problem are mapped into a characteristic function,the so-called spectral density g, while the optical behavior described by the function t. Theeffective dielectric function εeff can be written as:

εeff (ω) = ε1(ω)

(1− f

∫ 1

0

g(y)

t(ω)− ydy

)(6.43)

where

t =ε1(ω)

ε1(ω)− ε2(ω)(6.44)

If t takes real values between 0 and 1 at some frequency, the integrand denominatorvanishes and the integral diverges. This situation is called electrostatic resonance.

The function g only depends on the system topology and does not have anything todeal with the dielectric properties of the constituents, all contained in the function t. gcan be considered as a non-negative distribution function of resonances correspondingto the specific arrangement of the constituents. It is normalized:∫ 1

0

g(y)dy = 1 (6.45)

and in the case of statistical isotropic samples, the moment rule is:∫ 1

0

yg(y)dy =1− f

3(6.46)


Figure 6.8: Values of the real part of the t function for Silver (solid lines) and Gold (dashed lines)implanted in PDMS (εm = 2.7, [183]) calculated with data taken from Johnson and Christy [167](blue lines) and [175] (red lines).

The spectral density g has a noticeable influence on εeff if t is sufficiently close to theinterval [0, 1] on the real t-axis: in this case g can be obtained from measurements of εeff .If t is far away from [0, 1], εeff is not sensitive to the system topology and therefore mustbe calculated with simple mixing formulas (e.g. Maxwell Garnett or Bruggeman). Thiscondition, also referred as rigorous bounds on εeff , is satisfied for silver or gold nanopar-ticles implanted in PDMS: Figure6.8 shows the values assumed by the t function in thevisible range by using data from different authors [175, 167] for the metals dielectricfunctions, and remains always in the range [0, 1] between 350nm and 900nm.

Only few approaches have been made to explicitly and exactly calculate the spectraldensity g for realistic systems, like a dilute suspension of metal spheres [195], because ofthe Fredholm integral form of the expression for εeff . In the other cases in which g cannot be explicitly calculated, it can be obtained from the measurement of εeff in an itera-tive way since, in most cases, slight errors in εeff may cause enormous numerical errorsin g. The spectral density is thus adjusted to measurements using genetic algorithms.First an arbitrary starting function g, satisfying the normalization rules 6.45 and 6.46, isdiscretized. A random number generator assigns three points of the spectral density tobe changed keeping the first two moments conditions satisfied. If the spectral densitystill is non-negative, the new integrals are calculated from the old ones and the new spec-trum is compared with the measures. If the agreement is better than before, the iterationcontinues with the same variations. If not, different variations are tried. The resultingspectral density is finally smoothed using a momentum-conserving algorithm.

Bergman representation represents a powerful approach for the non-destructive in-vestigation of the arrangement of the nanoparticles inside the polymeric matrix by meansof optical measurements, mainly if the nanocomposite is subjected to a mechanical stress.Moreover it is worth noting that the same spectral density g obtained from optical mea-surements can be used to model other physical properties (always by using the Bergmanrepresentation), like the electrical or thermal conductivity. the Bergman representationwill not be used for the description of the experimental results obtained and reportedin the next chapter, but is a fundamental brick for the development of the future workderiving from the conclusions of this work.

CHAPTER 7

SPR in SCBI metal-polymer nanocomosites

The study of the plasmonic response of the Ag/PDMS nanocomposite synthesized bymeans of SCBI is driven by the necessity of investigating the optical properties of thematerial used for the fabrication of elastomeric mirrors or gratings and improve the fab-rication process. Different processes are involved in the evolution of the optical responseof the Ag/PDMS nanocomposite (PDMS swelling and nanoparticles oxidation, diffu-sion and aggregation among them) that must be investigated through nanocompositestreated with different processes. Nanoparticles oxidation was catalysed by thermal an-nealing at different temperatures and different times. The same treatments also enhancenanoparticles diffusion and aggregation in the polymer matrix. The deposition of a sec-ond PDMS layer after implantation was tried as a possible solution for the freezing ofthe nanocomposite state both in terms of oxidation and diffusion/aggregation.

Moreover, since elastomeric optical devices are often used in applications where de-formability is a requirement, the surface plasmon resonance was also characterized as afunction of sample elongation. Preliminary results will be presented at the end of thischapter, showing interesting results regarding the nanocomposite internal dynamics.These results also helps in better understanding the cluster topology and the electronictransport properties of such systems.

All the samples have been characterized by the acquisition of absorbance spectrain the uv-visible wavelength range (350-900nm) with a Jasco 7850 spectrophotometer.The sample holder of this instrument was modified by adding a home made computer-controlled motorized linear stage, enabling to measure different samples simultaneouslyand repeatedly at given time intervals. The temperature of the measurement chamberof the spectrophotometer was approximately 30◦C. A reference spectrum of the barePDMS was acquired before each measurement and the absorbance nullified at 900nm,far from the resonance peak, before mounting the nanocomposite samples.

7.1 Absorption with different equivalent thicknesses

In section 4.3.1 reflectance spectra of the Ag/PDMS nanocomposites with different im-planted equivalent thicknesses were presented, showing an interesting non-monotonetrend. Highly filled nanocomposites exhibit a lower reflectance respect to nanocom-posites with lower amounts of implanted nanoparticles, at odd with expectations. Re-flectance is just one of the three components of the optical response of a material, to-gether with transmittance and absorbance. The sum of these three quantities is uni-tary, therefore reflectance can be attenuated by absorbance or transmittance. Of the two,transmittance decreases with the amount of nanoparticles implanted in the polymer, be-cause of the well-known Lambert-Beer’s law. Absorbance thus remains the only possible

95

96 7.1 Absorption with different equivalent thicknesses

Figure 7.1: (a) Absorbance spectra of nanocomposites with different implanted equivalent thick-nesses (b) Plot of the baseline absorbance for different equivalent thicknesses. The baseline wascalculated as the mean absorbance far from the resonance peak, in the wavelength interval 850-900nm.

explanation for the reflectance behavior of metal-polymer nanocomposites with increas-ing implanted equivalent thicknesses.

Absorption spectra of samples implanted with different equivalent thicknesses (290,120, 60, 30nm, named c5, c6, c7 and c8 respectively) were acquired by uv-vis spec-troscopy and the results shown in Figure7.1a. By increasing the equivalent thicknessan enhancement of the absorbance is observed, as expected. However the whole spec-trum is shifted toward larger values of absorbance, at odd with the expectations of theMaxwell-Garnett model, in which the baseline (the region of spectrum far from the reso-nance, typically in the near-infrared in the current case) is expected to be null. The highbaseline of the reported spectra can be explained with the aggregation of a fraction of themetal nanoparticles inside the polymer, forming a buried bulk metal layer with a plasmonresonance shifted to the infrared region and causing a uniform absorbance in the wholewavelength visible range considered for the measurements. This simple model, alreadypresented in [118] in order to conciliate optical and electrical measurements, is schemat-ically reported in Figure7.2. An increase of the implanted equivalent thickness leads tothe growth both in density and thickness of this bulk layer, enhancing the absorbancethroughout the entire spectrum, as observed in Figure7.1. Additionally, as already seenin section 3.4.4 and in [22], a swelling of the PDMS is observed upon implantation. Thisphenomenon drags the non-aggregated metal nanoparticles far away from the bulk layer,giving rise to the formation of a polymer layer with dispersed nanoparticles above thebulk layer, responsible for the presence of the surface plasmon resonance.

Baseline levels were measured by averaging the infrared tails of spectra (in the regionbetween 850nm and 900nm), far from the resonance peaks. The results of this analysisfor the different values of equivalent thickness are reported in Figure7.1b. The increaseof absorbance with the amount of implanted nanoparticles justifies the non-monotonebehavior of the reflectance observed in Figure4.1b: the larger number of nanoparticles inthe polymer and the subsequent formation of thicker nanostructured bulk layers causean enhancement of absorbance, negatively affecting the reflectance of the material. Onthe opposite, reflectance of nanocomposites implanted with a low equivalent thicknessis limited by the low number of metal nanoparticles. The trade-off between high values

SPR in SCBI metal-polymer nanocomosites 97

Figure 7.2: (a) Schematic picture of the evolution of the metal-polymer nanocomposite with in-creasing implantation time and thus equivalent thickness, with the formation of a buried bulklayer. (b) Final configuration of the metal-polymer nanocomposite after implantation and poly-mer swelling: The upper region with dispersed nanoparticles is responsible for the surface plas-mon resonance peak in the absorption spectrum, while the buried bulk layer affects the level of thebaseline. (From [118]).

Figure 7.3: Absorbance spectra of samples c6, c7 and c8 with the subtracted baseline and withdifferent vertical scales for a direct comparison of the shape of the surface plasmon resonancepeaks.

of reflectance and limited absorbance is achieved for an equivalent thickness of approx-imately 60nm, as already discussed in section 4.3.1.

In order to analyze the dynamics of the isolated nanoparticles in the upper region ofthe nanocomposite, the baseline was subtracted from the spectra of the samples c6, c7and c8, and the results presented in Figure7.3. Sample c5 was not considered becauseof the high absorbance (near to the instrumental dynamic range) that avoids to obtaina smooth resonance peak from which useful and reliable information can be extracted.By subtracting the baseline, the contribution of the surface plasmon resonance is high-lighted and exhibits a double-peak structure, mainly for nanocomposites implanted withlower equivalent thicknesses. The first peak, at approximately 400nm, is due to the sur-face plasmon resonance of pure silver nanoparticles and its shape can be well comparedwith the Maxwell-Garnett simulated spectra (Figure6.5). The second peak, positioned be-tween 500nm and 600nm, is supposed to be caused by the surface plasmon resonance ofoxidized silver nanoparticles, describable with the shell-core model introduced in sec-tion 6.1.4 (Figure6.6).

The first peak, due to the pure silver nanoparticles, undergoes a blueshift with in-

98 7.2 Time evolution of the nanocomposite at RT

creasing implanted equivalent thickness, at odd with the Maxwell Garnett model andthe expected increase of the volume fraction. Similar results were observed and deeplystudied in Au/PDMS nanocomposites [21]. Blueshift can be explained with the decreaseof the filler volume fraction caused by aggregation of nanoparticles to the bulk layer andto the increase of the nanocomposite total volume following the swelling of the poly-mer, enhanced by larger implanted equivalent thicknesses (as observed and explainedin [22]).

More interesting, the behavior of the second peak due to the oxidized silver nanopar-ticles. Its relative intensity respect to the intensity of the first peak is lower and shifts to-ward longer wavelengths for larger implanted doses. According to the shell-core modelintroduced in section 6.1.4 this trend is compatible with thicker oxide shell around the sil-ver nanoparticles in the case of low implanted nanocomposite respect to high implanteddoses. As already stated in section 6.1.4, PDMS is permeable to oxygen and thus, in thecase of a low equivalent thickness, the few implanted nanoparticles are all exposed tooxygen and they therefore undergo oxidation. If a larger amount of filler is implanted,nanoparticles near the surface of the nanocomposite shield the underlying particles, thuslimiting their oxidation.

7.2 Time evolution of the nanocomposite at RT

Silver nanoparticles oxidation gives rise to a second resonance peak in the absorptionspectrum of the Ag/PDMS nanocomposite, also affecting the reflectance of the material.It is thus necessary to investigate and control the time evolution of the surface plasmonresonances, both at room temperature and after thermal annealing processes, in order toimprove the fabrication process of the elastomeric optical devices. Only nanocompositesimplanted with an equivalent thickness of 60nm were characterized.

Absorbance spectra of similar samples were acquired every 30 minutes up to 456hours after implantation. The samples were kept in standard conditions, at room tem-perature, between the measurements. Since all the samples show the same behavior,in Figure7.4a only the results relative to one sample is shown after certain significativetimes after implantation. The baseline was subtracted from spectra in order to directlycompare the shape of the resonance peaks with the theoretical models. The time evolu-tion of the intensity relative to the pure silver peak and to the baseline (calculated as theaverage of the spectrum between 850nm and 900nm) are reported in Figure7.4b, whilethe position of the pure silver peak in time is shown in Figure7.4c.

From Figure7.4a we can see that after 1 hour the peak due to the oxidized particlesarises and reaches a steady shape after 12 hours. The position of this second peak isdifficult to determine, because of the superposition with the plasmonic peak of the puresilver, that avoids to reliably distinguish the two contributes. In a similar way, the bulkabsorbance does not change significantly, meaning that at the measurements tempera-ture (approximately 30◦C) diffusion and aggregation of the silver nanoparticles to thebulk layer is limited. Diffusion and aggregation between the nanoparticles take place inthe upper swollen region, causing a redshift of the pure silver peak that changes from400nm to approximately 410-412nm after 144 hours and then remains stable.

Two sets of samples were measured at the same time after implantation, but one ofthem was passivated after 36 hours and left at room temperature (approximately 20◦C)for 96 hours for the PDMS passivation layer crosslinking. In order to compare the evo-lution data, also the first non-passivated set was kept in the same conditions, at a tem-perature lower than the measurement chamber. This change of temperature could havecaused a temporary rearrangement of the nanocomposite or a change in the swollen


Figure 7.4: (a) Absorbance spectra of the Ag/PDMS nanocomposite sample at eight significativetimes after implantation. (b) Plot of the absorbance versus time of the intensity of the pure silverpeak (the first peak) and of the baseline, calculated as the average of the spectrum in the range850nm-900nm. (c) First peak position in time

polymer density, affecting the optical response of that region of the sample. After havingresumed the measurements at 30◦C, the sample recovered the behavior before the passi-vation of the other set of samples. It is thus clear that the optical response of Ag/PDMSnanocomposites is highly sensitive to slight changes in the environmental temperature.This effect will be deeply investigated in the next section.

7.3 Thermal annealing

In order to investigate the effect of thermal annealing on the optical response of theAg/PDMS nanocomposites, a set of three samples were heated at different tempera-tures (150◦C, 100◦C, and 50◦C) for 1 hour and compared to a sample kept at room tem-perature, all implanted with an equivalent thickness of 60nm. The results, presentedin Figure7.5, show a different behavior between the two samples heated at 150◦C and100◦C, and the two samples heated at 50◦C and kept at room temperature. Samplesannealed at 100◦C and 150◦C present both the peaks due to pure silver and silver oxideparticles, while those annealed at 50◦C and kept at RT are characterized only by a peakaround 400nm due to pure silver nanoparticles, meaning that a temperature of 50◦C is

100 7.3 Thermal annealing

Figure 7.5: Spectra of samples immediately after annealing at different temperatures, indicated inthe legend (a), and 24 hours after annealing (b). In (c) and (d) the wavelength and the intensity ofthe pure silver resonance peak respectively versus the annealing temperature.

not sufficient to trigger an acceleration of the oxidation process. The intensity and po-sition of this first peak for the four samples were highlighted in Figure7.5c and d: withthe increase of the annealing temperature the peaks redshift and attenuate. This trendcan be explained through the nanocomposite model introduced in the previous sectionsand with a reduction of the thickness of the swollen region. The shrinking of the swollenpolymer region (already observed and measured in [118]) induces an increase of thevolume fraction (and thus a redshift) and an increase of the diffusion and aggregationbetween the nanoparticles and of the nanoparticles to the bulk layer, enhanced by thethermal energy of the annealing process. The reduction of the number of nanoparticlesin the swollen region causes a decrease of intensity of the first resonance peak while theaggregation of particles to the bulk layer an increase of the spectra baselines, as clearlyvisible in Figure7.5a.

24 hours after annealing the four samples were measured again: samples annealedat 100◦C and 150◦C are identical to the first measures, while those annealed at 50◦C andkept at RT show the arising of the peak due to silver oxide, while the baselines of the foursamples present no variations (as already observed in Figure7.4b). From these results itis clear that annealing processes at temperatures higher or equal to 100◦C acceleratethe oxidation of the silver nanoparticles and enhance the aggregation of nanoparticles


Figure 7.6: (a) Time evolution of the surface plasmon resonance of the Ag/PDMS nanocompositesystem: the baseline was subtracted in order to directly compare the shape of spectra. (b) Evolu-tion of the pure silver peak intensity and of the baseline level in time. (c) Time evolution of thepure silver peak position.

to the bulk layer, contributing to a flattening (and intensity increase) of the absorbancespectrum, thus affecting reflectance.

The high sensitivity of SCBI Ag/PDMS to temperature, due to the complex nanopar-ticles internal dynamics, deserves a deeper study of the time evolution of the nanocom-posite optical response under prolonged thermal annealing. A set of samples was an-nealed at 100◦C and measured each 20 minutes in order to investigate the evolution ofthe nanoparticles dynamics through surface plasmon resonances. All the samples be-have identically, so only the results and analyses relative to one sample are reported inFigure7.6 as representatives.

The silver oxide resonance peak arises after only 20 minutes of annealing at 100◦Cand its intensity is higher than the intensity of the pure silver nanoparticles peak, thatdecreases because of the strong oxidation reaction activated by the temperature. Theevolution shows a strong redshift of the silver oxide peak, accompanied by a decreasein its intensity, as predicted by the shell-core model with an increase of the oxide shellthickness, until its disappearing under the first peak.

Time evolution of the pure silver SPR peak intensity and of the spectrum baselinelevel are plotted in Figure7.6b: the attenuation of the SPR peak intensity is accompaniedby an increase of the spectrum baseline. This confirms the hypothesis according to which

102 7.4 Effect of passivation on SPR

thermal annealing improve the silver nanoparticles diffusion and aggregation to the bulkburied layer supposed in the SCBI nanocomposite model.

Also the time evolution of the position of the silver nanoparticles peak, shown inFigure7.6c, is interesting. An initial redshift is observed, caused by the nanoparticlesaggregation and shrinking of the swollen polymer [118]. After approximately 1 hour atrend inversion and a shift toward shorter wavelengths take place. This behavior can beexplained again with the aggregation of silver nanoparticles to the bulk layer togetherwith an adjustment of the polymer volume, causing a decrease of the swollen regionvolume fraction.

The samples were kept at room temperature for one week and then measured again,without any further significant variation in the absorbance spectra.

7.4 Effect of passivation on SPR

In section 7.2 and 7.3 we have seen that the nanocomposite optical properties changewith time and with thermal annealing because of the rearrangement and oxidation ofsilver nanoparticles and the swelling or shrinking of the superficial layer of the poly-meric substrate. A method for the freezing of the nanocomposite optical response in agiven state is required for the exploitation of the ns-Ag/PDMS system in practical appli-cations. A possible solution relies in the passivation of the nanocomposite, consisting inthe spin-coating of a second PDMS layer on the implanted nanocomposite. This processalso helps in improving the handling and reliability of the nanocomposite based device,since it protects the implanted surface from scratching or sweeping. Moreover, even ifPDMS is permeable to oxygen [187], a PDMS capping layer helps also limiting silvernanoparticles oxidation.

Samples in different evolution states (c10 and c11 annealed and c25 kept at roomtemperature) were passivated with a PDMS capping layer deposited by spin coatingat 1000rpm for 60 seconds and crosslinked at RT for 48 hours. Absorption spectra, ac-quired by uv-vis spectroscopy, are reported in Figure7.7a. Once again the baseline wassubtracted from the spectra in order to directly compare the features of the resonancepeaks.

The same trend is observed for all the three samples: both the pure silver and the ox-idized silver nanoparticles peaks shift toward shorter wavelengths, meaning a decreaseboth of the silver particles volume fraction according to the Maxwell Garnett model, andof the oxide shell thickness according to the shell-core model. Moreover, also the base-line level decreases, as shown in figure 7.7b. These processes can be explained by meansof two different mechanisms. Regarding the decrease of the volume fraction, it is wellknown that PDMS is highly sensitive to organic solvents present in the PDMS liquidprecursor spin-coated on the nanocomposite. Solvents are absorbed by the elastomericmatrix causing a swelling of the polymer [101, 102], an increase of the nanocompositetotal volume and thus a decrease of the nanoparticles volume fraction. The swelling ofthe nanocomposite polymeric matrix also drags silver nanoparticles away from the bulklayer, causing a decrease of the observed spectrum baseline level.

The hypothesis at the basis of the silver oxide particles blueshift is more complicated.The PDMS liquid precursor also containes silicone monomers that get cross-linked bya Platinum-complex based catalyzer [196, 197]. PDMS liquid precursor diffuse in thenanocomposite polymer matrix and both silver and silver oxide nanoparticles partici-pate in the crosslinking reaction, causing a de-oxidation of the latter. An interaction be-tween the polymer chains and metal nanoparticles was already observed for different


Figure 7.7: Absorption spectra of the nanocomposite samples before and after the passivation pro-cess. (b) Change of intensity of the silver SPR peak and of the spectra baseline level before andafter the passivation process.

systems [198, 199], in particular between silicon dioxide substrates and the oxygen ofunterminated PDMS chains, of which the liquid PDMS precursor is rich.

The same samples were kept at room temperature after the passivation process andmeasured again after one week, showing no significant modifications in the opticalresponse. Thus we can conclude that passivation affects the optical properties of thenanocomposite material but the new state achieved is freezed and does not undergo fur-ther modifications.

7.5 Stretching measurements

Since elastomeric metal-polymer nanocomposites synthesized by SCBI are supposed tobe exploited in stretchable or deformable optical devices like mirrors or gratings, a studyof the behavior of the optical response of the material subjected to mechanical stress isalso necessary. In this section very preliminary results regarding a Au/PDMS stretchednanocomposite system will be presented. Gold was chosen instead of Silver as nanocom-posite filling metal, thanks to its inertness to oxygen (it does not suffer oxidation likesilver) and because the expected surface plasmon resonance peak falls in the middle ofthe visible range, at approximately 550-600nm, thus large tunabilities can be observed.

104 7.5 Stretching measurements

Figure 7.8: (a) Absorbance spectra of stretched Au/PDMS nanocomposite. A clear redshift of theSPR peak position is evident. (b) SPR Peaks position and intensity plotted as a function of theapplied strain.

Moreover the behavior of the optical response of the Au/PDMS nanocomposite synthe-sized by SCBI is already known, since it was characterized by uv-vis spectroscopy [118]as a function of time, thermal annealing and implanted equivalent thickness.

A PDMS substrate was implanted by Supersonic Cluster Beam Implantation with anequivalent thickness of approximately 100nm and then mounted on the uv-vis spec-trophotometer modified sample holder. One extremity was clamped (with a Teflon-covered aluminum bar) in a support fixed to the sample holder, while the other sideclamped in a moving support, coupled with a computer controlled stepper motor. Onthe reference beam a freestanding bare PDMS layer serves as background for the mea-surements. The Au/PDMS nanocomposite was measured two weeks after the implan-tation process in order to limit the contribution of the SPR natural time evolution dueto the diffusion and aggregation of nanoparticles inside PDMS. The nanocomposite filmwas stretched up to 40% and absorbance spectra acquired at stretching intervals of 3%.The preliminary results are presented in Figure7.8.

During a single stretching cycle a redshift and an attenuation of the SPR Peak is ob-served. According to Maxwell Garnett theory, a redshift means an increase of the volumefraction and this can be caused by two phenomena: a rearrangement and aggregation ofthe nanoparticles driven by the stretching of the polymeric chains, and/or a decreaseof the nanocomposite total volume because of the shrinking of the polymer in the two

Conclusions and perspectives 105

directions orthogonal to the stretching direction, due to the Poisson ratio (that is 0.5 [200]for PDMS). However an attenuation of the SPR peak intensity is observed, at odd withthe expectations of the Maxwell Garnett model. This decrease could be explained withan aggregation of the gold nanoparticles to a bulk layer, similarly to the silver case seenin the previous sections, but in this case the baseline can not be determined since goldexhibits a much larger SPR peak, covering a range between 500nm and 900nm. A secondhypothesis for the SPR peak attenuation relies on the formation of aggregates orientedin the stretching direction. According to this assumption the same number of nanopar-ticles gives rise to a lower number of linear aggregates with a high aspect ratio, shiftingthe SPR toward longer wavelengths, as observed in [201]. Nanoparticles aggregationin structures with larger aspect ratio even without applied stretching was supposed bythe comparison of the measured nanocomposite surface Young modulus with theoreticalexpectations of Guth model seen earlier in section 3.4.3. The formation of nanoparticlesaggregates in the stretching direction in deformed metal-polymer nanocomposites wasinvestigated in [202] and is at the basis for the explanation of the improvement of theelectrical conductance upon stretching measured in [21, 118].

The rearrangement of the nanoparticles inside the Au/PDMS nanocomposite in thestretching direction may also linearly polarize the light passing through the nanocompos-ite. This effect can be exploited for the direct investigation of the aggregation state of theimplanted nanoparticles with uv-vis spectroscopy, measuring the intensity attenuationof transmitted incident light, polarized at different angles by means of a revolving linearpolarizer, both in the measure and in the reference beams.

The nanoparticles aggregation status can be also investigated applying the Bergmanmodel, introduced in section 6.2, to the obtained uv-vis absorption spectra. The spectraldensity g obtained from the Bergman model gives information on the distribution ofresonances and thus to the structure of the nanoparticle aggregates inside the polymer.However, since the function g is obtained from a fitting procedure of the spectra, a non-trivial genetic algorithm and computational effort are required.

The knowledge of the nanoparticles structure and its effect on the optical propertiesof the nanocomposite is of great interest for the fabrication and tuning of the elastomericoptical devices characteristics, but serves also for a better understanding of the electronictransport mechanism of such materials.

Conclusions and perspectives

In this thesis I demonstrated the possibility to fabricate reliable nanocomposite-basedelastomeric optical components working in reflection by means of Supersonic ClusterBeam Implantation. Unlike reflective elastomeric optical components fabricated by clas-sical techniques the resilience of the metal-elastomer nanocomposite ensures a sufficientelasticity of the reflective layer while maintaining an extremely low surface smooth-ness, limiting the fraction of scattered light. The superior optical and mechanical prop-erties of these devices compared to currently available elastomer metallization tech-niques allowed to design and to build cheaper and simpler novel optical instrumentslike monochromators and spectrometers, characterized by an easier and linear wave-length selection respect to currently available commercial spectrophotometers. In partic-ular SCBI elastomeric optical components applied to arbitrarily shaped and non-opticalgrade surfaces can replace and add new features to existing concave diffraction gratings,opening the way to the use of our elastomeric components in novel cheaper and easieroptical setups and instruments.

The optical properties of elastomeric optical devices can be finely tuned during thenanocomposite synthesis and with post-implantation thermal treatments, as shown bythe study of the behavior of Surface Plasmon Resonance observed in the nanocompositeabsorption spectra. In particular I demonstrated that passivation of the nanocompositelayer helps in preventing silver nanoparticles oxidation and mobility, and improving thehandling of the device. Preliminary results on the optical behavior of the nanocompositeupon stretching, suggesting a re-arrangement of the nanoparticles inside the polymer,are particularly encouraging in view of a better understanding both of the behavior ofelastomeric tunable gratings or mirrors.

The results obtained in this work open new research opportunities and possible ap-plications of elastomeric optical devices. Among these applicative fields deformable mir-rors for small adaptive optics systems and hyperspectral imaging systems, exploiting theresilience of elastomeric mirrors and the imaging capabilities of arbitrarily shaped elas-tomeric diffraction gratings respectively, are the most promising.

Reflective membranes, actuated by electrostatic forces, for cheap and reliable com-pact adaptive optics systems can be easily fabricated by means of SCBI Ag/PDMS nano-composites. The reflectivity of our nanocomposites can be combined with the possibilityto create conductive nanocomposite micropatterns [21, 22] by means of SCBI with stencilmasks without any deterioration of the elastomer. The result would be a single reflectiveelastomeric device comprising patterned electrodes, representing the ideal candidate forthe replacement of currently available continuous faceplates mirrors, affected by a highfragility and a complex (and expensive) piezoelectric-based actuation system.

108 Bibliography

Finally, reflective elastomeric nanocomposite-based diffraction gratings synthesizedby SCBI may represent a significative breakthrough, making hysperspectral imaging sys-tems commonly available and useful in everyday life. Our elastomeric devices wouldovercome the limited focusing and imaging capabilities of existing diffraction gratings.The high fabrication cost of concave gratings or the large dimensions and costs of alter-native mounts, based on planar gratings and complex focusing and corrective systems,limit the use of hyperspectral imaging in niche applications [203, 204]. Novel cheaper,simpler and more compact hyperspectral imaging systems can be developed thanks tothe results relative to the applications of nanocomposite-based elastomeric gratings to ar-bitrarily shaped non optical grade surfaces. Images can be thus diffracted, focused andoff-axis aberrations corrected by a single inexpensive elastomeric optical component.

Bibliography

[1] H. W. Babcock. The possibility of compensating astronomical seeing. Pub. Astr.Soc. Pac., 65:229, 1953.

[2] A. Buffington, F. S. Crawford, R. A. Muller, A. J. Schemin, and R. G. Smits. J. Opt.Soc. Am., 67:298, 1977.

[3] A. Buffington, F. S. Crawford, R. A. Muller, and C. D. Orth. J. Opt. Soc. Am., 67:304,1977.

[4] J. W. Hardy, J. E. Lefebvre, and C. L. Koliopoulos. Real-time atmospheric compen-stion. J. Opt. Soc. Am., 67:360, 1977.

[5] P. Wizinowich, D. S. Acton, C. Shelton, P. Stomski, J. Gathright, K. Ho, W. Lupton,K. Tsubota, O. Lai, C. Max, J. Brase, J. An, K. Avicola, S. Olivier, D. Gavel, , B. Mac-intosh, A. Ghez, and J. Larkin. First light adaptive optics images from the keckii telescope: A new era of high angular resolution imagery. Pub. Astr. Soc. Pac.,112:315, 2000.

[6] R. B. Dunn. Proc. SPIE, 1271:216, 1990.[7] R. C. Smithson. Proc. SPIE, 779:18, 1987.[8] R. C. Smithson, M. L. Peri, and R. S. Benson. Appl. Opt., 27:1615, 1988.[9] F. Merkle, K. Freischlad, and H. L. Reischmann. Deformable mirror with combined

piezoelectric and electrostatic actuators. Proc. SPIE, 332:260, 1982.[10] O. Solgaard, F. S. A. Sandejas, and D. M. Bloom. Deformable grating optical mod-

ulator. Opt. Lett., 17:688, 1992.[11] Y. T. Yu, W. Z. Yuan, T. P. Li, and B. Yan. Develpment of a micromechanical pitch-

tunable grating with reflective/transmissive dual working modes. J. Micromech.Microeng., 20:065002, 2010.

[12] J. L. Wilbur, R. J. Jackman, and G. M. Whitesides. Elastomeric optics. Chem. Mater.,8:1380, 1996.

[13] T. Bifano. Mems deformable mirrors. Nat. Photonics, 5:21, 2011.[14] N. J. Douville, Z. Li, S. Takayama, and M. D. Thouless. Fracture of metal coated

elastomers. Soft Matter, 7:6493, 2011.[15] Y. Shao and M. A. Brook. Structured metal films on silicone elastomers. J. Mater.

Chem., 20:8548, 2010.[16] G. Ouyang. Hybrid Integrated Components for Optical Display Systems - Electroactive

Polymer Based Micro-Opto-Electro-Mechanical Actuators. PhD thesis, Department ofPhysics, University of Oslo, 2012.

[17] W. Caseri. Nanocomposites of polymers and metals or semiconductors: Historicalbackground and optical properties. Macromol. Rapid Commun., 21:705, 2000.

109

110 Bibliography

[18] K. R. Berry Jr, A. G. Russell, P. A. Blake, and D. K. Roper. Gold nanoparticles re-duced in situ and dispersed in polymer thin films: optical and thermal properties.Nanotechnology, 23:375703, 2012.

[19] S. Rosset. Metal Ion Implanted Electrodes for Dielectric Elastomer Actuators. PhDthesis, Programme doctoral en microsystemes et microelectronique - EPFL, 2009.

[20] M. Niklaus, S. Rosset, P. Dubois, M. Dadras, and H. Shea. Modification of con-ductivity and of mechanical properties of electroactive polymer (eap) thin filmsby titanium ion implantation. In MRS Fall Meeting 2007, 2007.

[21] G. Corbelli, C. Ghisleri, M. Marelli, P. Milani, and L. Ravagnan. Highly-deformablenanostructured elastomeric electrodes with improving conductivity upon cyclicalstretching. Adv. Mater., 23(39):4504, 2011.

[22] C. Ghisleri, F. Borghi, L. Ravagnan, A. Podesta, C. Melis, L. Colombo, and P. Mi-lani. Patterning of gold-polydimethylsiloxane (au-pdms) nanocomposites by su-personic cluster beam implantation. J. Phys. D: Appl. Phys., 47:015301, 2014.

[23] C. Ghisleri, M. Siano, L. Ravagnan, M. A. C. Potenza, and P. Milani.Nanocomposite-based stretchable optics. Laser & Photon. Rev., 7:1020, 2013.

[24] C. Ghisleri, A. Annoni, S. Capra, M. A. C. Potenza, L. Ravagnan, and P. Milani.A simple scanning spectrometer based on a stretchable reflective grating. AppliedPhysics Letters, Submitted, 2013.

[25] M. Loktev, O. Soloviev, and G. Vdovin, editors. OKO Technologies, 2008.[26] Multiple Authors. Adaptive Optics for Vision Science: principles, practices, design and

applications. Wiley-Interscience, 2006.[27] J. M. Beckers. Adaptive optics for astronomy: Principles, performance, and appli-

cations. Annu. Rev. Astron. Astrophys., 31:13, 1993.[28] N. R. Zelmanovitch. The outreach plan around the ?rst science with the gran tele-

scopio canarias (gtc). Proceedings IAU Symposium, 260:1, 2009.[29] W. G. Hulburd. Segmented mirrors for atmospheric compensation. Proc. SPIE,

1130:42, 1989.[30] A. Roorda. Adaptive optics ophthalmoscopy. J. Refract. Surg., 16:S602, 2000.[31] EASO. The 4m international liquid mirror telescope project -

http://www.aeos.ulg.ac.be/lmt/index.php.[32] P. Laird, N. Caron, M. Rioux, E. F. Borra, and A. Ritcey. Ferrofluidic adaptive

mirrors. Appl. Opt., 45:3495, 2006.[33] D. Brousseau, E. F. Borra, and S. Thibault. Wavefront correction with a 37-actuator

ferrofluid deformable mirror. Opt. Express, 15:18190, 2007.[34] A. Iqbal and F. B. Amara. Modeling of a magnetic-fluid deformable mirror for

retinal imaging adaptive optics systems. Int. J. Optomech., 1:180, 2007.[35] A. Iqbal and F. B. Amara. Modeling and experimental evaluation of a circular

magnetic-fluid deformable mirror. Int. J. Optomech., 2:126, 2008.[36] James webb space telescope (jwst) indepent comprehensive review panel (icrp)

final report. Technical report, NASA, 2012.[37] T. Young. On the mechanism of the eye. Phil. Trans. R. Soc. Lond., 91:23, 1801.[38] H. Helmoltz. Physiological Optics. Optical Society of America, Rochester (NY),

1924.[39] M. S. Smirnov. Measurement of the wave aberration of the human eye. Biofizika,

6:687, 1961.[40] A. W. Dreher, J. F. Bille, and R. N. Weinreb. Active optical depth resolution im-

provement of the laser tomographic scanner. Appl. Opt., 28:804, 1989.

Bibliogrphy 111

[41] J. Liang, B. Grimm, S. Goelz, and J. F. Bille. Objective measurement of wave aber-rations of the human eye with the use of a hartmann-shack wave-front sensor. J.Opt. Soc. Am. A, 11:1949, 1994.

[42] J. Liang and D. R. Williams. Aberrations and retinal image quality of the normalhuman eye. J. Opt. Soc. Am. A, 14:2873, 1997.

[43] H. Hofer, P. Artal, B. Singer, J. L. Aragon, and D. R. Williams. Dynamics af theeye’s aberrations. J. Opt. Soc. Am. A, 18:497, 2001.

[44] J. Caroll, D. C. Gray, A. Roorda, and D. R. Williams. Recent advances in retinalimaging with adaptive optics. Opt. Photon. News, 16:36, 2005.

[45] A Roorda and D. R. Williams. The arrangement of the three cone classes in theliving human eye. Nature, 397:520, 1999.

[46] M. J. Booth. Adaptive optics in microscopy. Phil. Trans. R. Soc. Lond. A, 365:2829,2007.

[47] O. Albert, L. Sherman, G. Mourou, T. B. Norris, and G. Vdovin. Smart microscope:an adaptive optics learning system for aberration correction in multiphoton con-focal microscopy. Opt. Lett., 25:52, 2000.

[48] M. J. Booth, M. A. A. Neil, R. Juskaitis, and T. Wilson. Adaptive aberration correc-tion in a confocal microscope. Proc. Natl Acad. Sci. USA, 99:5788, 2002.

[49] B. F. Grewe, F. F. Voigt, M. van ’t Hoff, and F. Helmchen. Fast two-layer two-photon imaging of neuronal cell populations using an electrically tunable lens.Biomed. Opt. Express, 2:2035, 2011.

[50] F. Druon, G. Cheriaux, J. Faure, J. Nees, M. Nantel, A. Maksimchuk, G. Mourou,J. C. Chanteloup, and G. Vdovin. Wave-front correction of femtosecond terawattlasers by deformable mirrors. Opt. Lett., 23:1043, 1998.

[51] W. Zeller, L. Naehle, P. Fuchs, F. Gerschuetz, L. Hildebrandt, and J. Koeth. Dfblasers between 760nm and 16µm for sensing applications. Sensors, 10:2492, 2010.

[52] J. Aleman, A. V. Chadwick, J. He, M. Hess, K. Horie, R. G. Jones, P. Kratochvıl,I. Meisel, I. Mita, G. Moad, S. Penczek, and R. F. T. Stepto. Definitions of terms re-lating to the structure and processing of sols, gels, networks, and inorganic-organichybrid materials. Pure Appl. Chem., 79:1801, 2007.

[53] D. Qin, Y. Xia, and G. M. Whitesides. Elastomeric light valves. Adv. Mater., 9:407,1997.

[54] C. Yang, K. Shi, P. Edwards, and Z. Liu. Demonstration of a pdms based hybridgrating and fresnel lens (g-fresnel) device. Opt. Express, 18:23529, 2010.

[55] A. N. Simonov, O. Akhzar-Mehr, and G. Vdovin. Light scanner based on a vis-coelastic stretchable grating. Opt. Lett., 30:949, 2005.

[56] A. N. Simonov, S. Grabarnik, and G. Vdovin. Stregrating diffraction gratings forspectrometry. Opt. Express, 15:9784, 2007.

[57] S. Rosset, B. M. OBrien, T. Gisby, D. Xu, H. R. Shea, and I. A. Anderson. Self-sensing dielectric elastomer actuators in closed-loop operation. Smart Mater.Struct., 22:104018, 2013.

[58] S. C. Truxal, K. Kurabayashi, and Y. C. Tung. Design of a mems tunable polymergrating for single detector spectroscopy. Int. J. Optomech., 2:75, 2008.

[59] B. A. Grzybowski, S. T. Brittain, and G. M. Whitesides. Thermally actuated in-terferometric ssensor based on the thermal expansion of transparent elastomericmedia. Rev. Sci. Instrum., 70:2031, 1999.

[60] G. Ye and X. Wang. Polymer diffraction gratings on stimuli-responsive hydrogelssurface: Soft-litographic fabrication and optical sensing properties. Sens. Actua-

112 Bibliography

tors, B, 147:707, 2010.[61] Patent us 7453646 b2 - fluidic adaptive lens systems and methods, 2008.[62] M. C. Hutley. Diffraction gratings. Academic Press, 1982.[63] C. Palmer. Diffraction Grating Handbook. Newport Corporation, 2005.[64] G. Y. Tian, R. S. Lu, and D. Gledhill. Surface measurement using active vision and

light scattering. Opt. Laser Eng., 45:131, 2007.[65] H. E. Bennett and J. O. Porteus. Relation between surface roughness and specular

reflectance at normal incidence. Journal of the Optical Society of America, 51:123,1961.

[66] P. Beckmann and A. Spizzichino. The Scattering of Electromagnetic Waves from RoughSurfaces. Pergamon, Oxford, 1965.

[67] T. Li and Z. Suo. Ductility of thin metal films on polymer substrates modulated byinterfacial adhesion. Int. J. Solids Struct., 44:1696, 2007.

[68] T. Li, Z. Huang, Z. Suo, S. P. Lacour, and S. Wagner. Stretchability of thin metalfilms on elastomer substrates. Appl. Phys. Lett., 85:3435, 2004.

[69] H. Huang and F. Spaepen. Tensile testing od free-standing cu, ag and al thin filmsand ag/cu multilayers. Acta mater., 48:3261, 2000.

[70] S. L. Chiu and P. S. Ho. Fracture of metalpolymer line structure. i. semiflexiblepolymide. J. Appl. Phys., 76:5136, 1994.

[71] Y. Xiang, T. Li, Z. Suo, and J. J. Vlaassak. High ductility of a metal film adherenton a polymer substrate. Appl. Phys. Lett., 87:161910, 2005.

[72] S. P. Lacour, D. Chan, S. Wagner, T. Li, and Z. Suo. Mechanisms of reversiblestretchability of thin metal films on elastomeric substrates. Appl. Phys. Lett.,88:204103, 2006.

[73] B. Budiansky and R. J. O’connell. Elastic moduli of a cracked solid. Int. J. SolidsStruct., 12:81, 1976.

[74] N. Laws and J. R. Brokenbrough. The effect of micro-crack systems on the loss ofstiffness of brittle solids. Int. J. Solids Struct., 23:1247, 1987.

[75] M. R. Begley, O. N. Scott, M. Utz, and H. Bart-Smith. Fracture of nanoscale copperfilms on elastomer substrates. Appl. Phys. Lett., 95:231914, 2009.

[76] S. P. Lacour, S. Wagner, Z. Huang, and Z. Suo. Stretchable gold conductors onelastomeric substrates. Appl. Phys. Lett., 82:2404, 2003.

[77] I. M. Graz, D. P. J. Cotton, and S. P. Lacour. Extended cyclic uniaxial loading ofstretchable gold thin-films on elastomeric substrates. Appl. Phys. Lett., 94:071902,2009.

[78] S. Olcum, A. Kocabas, G. Ertas, A. Atalar, and A. Aydinli. Tunable surface plasmonresonance on an elastomeric substrate. Opt. Express, 17:8542, 2009.

[79] N. Bowden, S. Brittain, A. G. Evans, J. W. Hutchinson, and G. M. Whitesides. Spon-taneous formation of ordered structures in thin flms of metals supported on anelastomeric polymer. Nature, 393:146, 1998.

[80] J. Y. Chung, A. J. Nolte, and C. M. Stafford. Surface wrinkling: A versatile platformfor measuring thin-film properties. Adv. Mater., 23:349, 2011.

[81] R. A. Guerrero, J. T. Barretto, J. L. V. Uy, I. B. Culaba, and B. O. Chan. Effectsof spontaneous surface buckling on the diffraction performance of an au-coatedelastomeric grating. Opt. Commun., 270:1, 2007.

[82] M. Aschwanden and A. Stemmer. Polymeric, electrically tunable diffraction grat-ing based on artificial muscles. Opt. Lett., 31:2610, 2006.

[83] N. Galler, H. Ditlbacher, B. Steinberger, A. Hohenau, M. Dansaschmuller,

Bibliogrphy 113

F. Camacho-Gonzales, S. Bauer, J. R. Krenn, A. Leitner, and F. R. Ausseneg. Electri-cally actuated elastomers for electro-optical modulators. Appl. Phys. B, 85:7, 2006.

[84] K. Wetzig and C. M. Shneider. Metal based thin films for electronics. Wiley-VCHGmbH and Co. KGaA, Weinheim, 2003.

[85] M. Kolle, B. Zheng, N. Gibbons, J. J. Baumberg, and U. Steiner. Stretch-tuneabledielectric mirrors and optical microcavities. Opt. Express, 18:4356, 2010.

[86] S. P. Lacour, J. Jones, Z. Suo, and S. Wagner. Design and performance of thin metalfilm interconnects for skin-like electronic circuits. IEEE Electr. Device Lett., 25:179,2004.

[87] Y. W. Jiang, F. T. Chuang, H. H. Chen, and S. C. Lee. Self-assembled ripplingstructure based on metal-elastomer composite for tunable plasmonics. IEEE Phot.Technol. Lett., 23:670, 2011.

[88] E. Guth. Theory of filler reinforcement. J. Appl. Phys., 16:20, 1945.[89] Y. P. Wu, Q. X. Jia, D. S. Yu, and L. Q. Zhang. Modeling young’s modulus of rubber-

clay nanocomposites using composite theories. Polym. Testing, 23:903, 2004.[90] L. Nicolais and G. Carotenuto. Metal-polymer nanocomposites. Wiley, 2005.[91] A. B. R. Mayer. Formation of noble metal nanoparticles within a polymeric matrix:

nanoparticle features and overall morphologies. Mater. Sci. Eng. C, 6:155, 1998.[92] F. Faupel, V. Zaporojtchenko, H. Greve, U. Schurmann, V. S. K. Chakravadhanula,

Ch. Hanish, A. Kulkarni, A. Gerber, E. Quandt, and R. Podschun. Deposition ofnanocomposites by plasmas. Contrib. Plasma Phys., 47:537, 2007.

[93] V. Zaporojtchenko, R. Podschun, U. Schurmann, A. Kulkarni, and F. Fau-pel. Physico-chemical and antimicrobial properties of co-sputtered ag/au-ptfenanocomposite coatings. Nanotechnology, 17:4904, 2006.

[94] C. V. Bechtolsheim, V. Zaporojtchenko, and F. Faupel. Influence of thermal treat-ment on the morphology and adhesion of gold films on trimethylcyclohexanepoly-carbonate. Appl. Phys. Surf., 151:119, 1999.

[95] D. V. Gioia. Preparation and formation mechanisms of uniform metallic particlesin homogeneous solutions. J. Mater. Chem., 14:451, 2004.

[96] M. Iwamoto, K. Kuroda, V. Zaporojtchenko, S. Hayashi, and F. Faupel. Productionof gold nanoparticles-polymer composite by quite simple method. Eur. Phys. J. D,24:365, 2003.

[97] D. Bloor, K. Donnelly, P. J. Hands, P. Laughlin, and D. Lussey. A metal-polymercomposite with unusual properties. J. Phys. D: Appl. Phys., 38:2851, 2005.

[98] Z. Lu, G. Liu, and S. Duncan. Poly(2-hydroxyethyl acrylate-co-methyl acry-late)/sio2/tio2 hybrid membranes. J. Membr. Sci., 221:113, 2003.

[99] R. A. Zoppi, S. das Neves, and S. P. Nunes. Hybrid films of poly(ethylene oxide-b-amide-6) containing sol-gel silicon or titanium oxide as inorganic fillers: effectof morphology and mechanical properties on gas permeability. Polymer, 41:5461,2000.

[100] U. Kreibig and M. Vollmer. Optical Properties of Metal Clusters. Springer, 1995.[101] J. N. Lee, C. Park, and G. M. Whitesides. Solvent compatibility of

poly(dimethylsiloxane)-based microfluidic devices. Anal. Chem., 75:6544, 2003.[102] A. Mata, A. J. Fleischman, and S. Roy. Characterization of polydimethylsiloxane

(pdms) properties for biomedical micro/nanosystems. Biomed. Microdevices, 7:281,2005.

[103] R. Ruggerone. Highly Filled Polymer Nanocomposite Films Derived from NovelNanostructured Latexes. PhD thesis, Programme doctoral en science et genie des

114 Bibliography

materiaux., 2009.[104] D. Fragiadakis, P. Pissis, and L. Bokobza. Glass transition and molecular dynamics

in poly(dimethylsiloxane)/silica nanocomposites. Polymer, 46:6001, 2005.[105] H. Zou, S. Wu, and J. Shen. Polymer/silica nanocomposites: Preparation, charac-

terization, properties, and applications. Chem. Rev., 108:3893, 2008.[106] S. Rosset and H. R. Shea. Flexible and stretchable electrodes for dielectric elastomer

actuators. Appl. Phys. A, 110:281, 2013.[107] P. Dubois, S. Rosset, S. Koster, J. Stauffer, S. Mikhaılov, M. Dadras, N. de Rooij, and

H. Shea. Microactuators based on ion implanted dielectric electroactive polymer(eap) membranes. Sens. Actuators, A, 130-131:147, 2006.

[108] S. Rosset, M. Niklaus, P. Dubois, and H. R. Shea. Metal ion implantation for thefabrication of stretchable electrodes on elastomers. Adv. Funct. Mater., 19:470, 2009.

[109] A. L. Stepanov, D. E. Hole, and P. D. Townsed. Reflectance of the dielectric layerscontaining metal nanoparticles formed by ion implantation. J. Non-Cryst. Solids,244:275, 1999.

[110] A. L. Stepanov. Synthesis of silver nanoparticles in dielectric matrix by ion im-plantation: a review. Rev. Adv. Mater. Sci., 26:1, 2010.

[111] M. Niklaus, S. Rosset, and H. Shea. Array of lenses with individually tunablefocal-length based on transparent ion-implanted eaps. Proc. SPIE, 7642:1, 2010.

[112] T. K. Shih, C. F. Chen, J. R. Ho, and F. T. Chuang. Fabrication of pdms (poly-dimethylsiloxane) microlens and diffuser using replica molding. Microelectron.Eng., 83:2499, 2006.

[113] P. J. Flory. Molecular theory of rubber elasticity. Polym. J., 17:1, 1985.[114] L. Ravagnan, G. Divitini, S. Rebasti, M. Marelli, P. Piseri., and P. Milani.

Poly(methyl methacrylate)-palladium clusters nanocomposite formation by su-personic cluster beam deposition: a method for microstructured metallization ofpolymer surfaces. J. Phys. D: Appl. Phys., 42:082002, 2009.

[115] M. Marelli, G. Divitini, C. Collini, L. Ravagnan, G. Corbelli, C. Ghisleri, A. Gi-anfelice, C. Lenardi, P. Milani, and L. Lorenzelli. Flexible and biocompatible mi-croelectrode arrays fabricated by supersonic cluster beam deposition on su-8. J.Micromech. Microeng., 21:045013, 2011.

[116] K. Wegner, P. Piseri, H. V. Tafreshi, and P. Milani. Cluster beam deposition: a toolfor nanoscale science and technology. J. Phys. D: Appl. Phys., 39:R439, 2006.

[117] H. Haberland, Z. Insepov, and M. Moseler. Molecular-dynamics simulation ofthin-film growth by energetic cluster impact. Phys. Rev. B, 51:11061, 1995.

[118] G. Corbelli. Synthesis and characterization of metal-polymer nanocomposites for stretch-able electronics applications. PhD thesis, Doctorate School in Physics, Astrophysicsand Applied Physics - University of Milan, 2012.

[119] G. Bongiorno, A. Podesta, L. Ravagnan, P. Piseri, P. Milani, C. Lenardi, S. Miglio,M. Bruzzi, and C. Ducati. Electronic properties and applications of cluster-assembled carbon films. J. Mater. Sci.: Mater Electron, 17:427, 2006.

[120] E. Barborini, G. Corbelli, G. Bertolini, P. Repetto, M. Leccardi, S. Vinati, and P. Mi-lani. The influence of nanoscale morphology on the resistivity of cluster-assemblednanostructured metallic thin films. New J. Phys., 12:073001, 2010.

[121] L. Diederich, E. Barborini, P. Piseri, A. Podesta, P. Milani, A. Schneuwly, andR. Gallay. Supercapacitors based on nanostructured carbon electrodes grown bycluster-beam deposition. Appl. Phys. Lett., 75:2662, 1999.

[122] L. G. Bettini, G. Bardizza, A. Podesta, P. Milani, and P. Piseri. Electrochemical

Bibliogrphy 115

impedance spectroscopy on nanostructured carbon electrodes grown by super-sonic cluster beam deposition. J. Nanopart. Res., 15:1429, 2013.

[123] M. Bruzzi, S. Miglio, M. Scaringella, G. Bongiorno, P. Piseri, A. Podesta, and P. Mi-lani. First study of humidity sensors based on nanostructured carbon films pro-duced by supersonic cluster beam deposition. Sens. Actuators, B, 100:173, 2004.

[124] T. Mazza, E. Barborini, I. N. Kholmanov, P. Piseri, G. Bongiorno, S. Vinati, P. Milani,C. Ducati, D. Cattaneo, A. Li Bassi, C. E. Bottani, A. M. Taurino, and P. Siciliano.Libraries of cluster-assembled titania films for chemical sensing. Appl. Phys. Lett.,87:103108, 2005.

[125] A. Podesta, G. Bongiorno, P. E. Scopelliti, S. Bovio, P. Milani, C. Semprebon, andG. Mistura. Cluster-assembled nanostructured titanium oxide films with tailoredwettability. J. Phys. Chem. C, 113:18264, 2009.

[126] R. Carbone, I. Marangi, A. Zanardi, L. Giorgetti, E. Chierici, G. Berlanda,A. Podesta, F. Fiorentini, G. Bongiorno, P. Piseri, P. G. Pellicci, and P. Milani. Bio-compatibility of cluster-assembled nanostructured tio2 with primary and cancercells. Biomaterials, 27:3221, 2006.

[127] P. Milani and S. Iannotta. Cluster Beam Synthesis of Nanostructured Materials.Springer, 1999.

[128] H. V. Tafreshi, G. Benedek, P. Piseri, S. Vinati, E. Barborini, and P. Milani. Aero-dynamic focusing of clusters into a high intensity and low divergence supersonicbeam. Europ. Phys. J. Appl. Phys., 16:149, 2001.

[129] H. V. Tafreshi, G. Benedek, P. Piseri, S. Vinati, E. Barborini, and P. Milani. A simplenozzle configuration for the production of low divergence supersonic cluster beamby aerodynamic focusing. Aerosol Sci. Technol., 36:593, 2002.

[130] Y. Yamamura and T. Muramoto. Depth profiles due to big cluster impacts. Phys.Rev. Lett., 69:1463, 1992.

[131] C. Anders and H. M. Urbassek. Dependence of cluster ranges on target cohesiveenergy: Molecular-dynamics study of energetic au402 cluster impacts. Nucl. In-strum. Methods Phys. Res., Sect. B, 266:44, 2008.

[132] S. J. Carroll, P. D. Nellist, and R. E. Palmer. Shallow implantation of ”size-selected”ag clusters into graphite. Phys. Rev. Lett., 84:2654, 2000.

[133] V. N. Popok, I. I. Azarko, R. I. Khaibullin, A. L. Stepanov, V. Hnatowicz, A. Mack-ova, and S. V. Prasalovich. Radiation-induced change of polyimide properties un-der high-fluence and high ion current density implantation. Appl. Phys. A, 78:1067,2004.

[134] R. Cardia, C. Melis, and L. Colombo. Neutral-cluster implantation in polymers bycomputer experiments. J. Appl. Phys., 113:224307, 2013.

[135] E. Barborini, P. Piseri, and P. Milani. A pulsed microplasma source of high intensitysupersonic carbon cluster beams. J. Phys. D: Appl. Phys., 32:L105, 1999.

[136] P. Milani and W. A. deHeer. Improved pulsed laser vaporization source for pro-duction of intense beams of neutral and ionized clusters. Rev. Sci. Instrum., 61:1835,1990.

[137] P. R. Willmott and J. R. Huber. Pulsed laser vaporization and deposition. Rev. Mod.Phys., 72:315, 2000.

[138] H. R. Siekmann, Ch. Luder, J. Faehrmann, H. O. Lutz, and K. H. Meiwes-Broer.The pulsed arc cluster ion source (pacis). Z. Phys. D, 20:417, 1991.

[139] NIST. Persistent lines of neutral mercury ( hg i ) - http://physics.nist.gov/.[140] S. Prahl. Rhodamine b - http://omlc.ogi.edu/spectra/photochemcad/index.html.

116 Bibliography

[141] C. Kittel. Introduction to Solid State Physics, 8th Edition. John Wiley & Sons, 2005.[142] N. W. Ashcroft and N. D. Mermin. Solid State Physics. Holt Rinehart & Winston,

1976.[143] P. Drude. Zur elektronentheorie der metalle. Annalen der Physik, 306:566, 1900.[144] J. N. Anker, W. P. Hall, O. Lyandres, N. C. Shah, J. Zhao, and R. P. van Duyne.

Biosensing with plasmonic nanosensors. Nat. Mater., 7:442, 2008.[145] J. Homola. Present and future of surface plasmon resonance biosensors. Anal.

Bioanal. Chem., 377:528, 2003.[146] X. Huang, I. H. El-Sayed, W. Qian, and M. A. El-Sayed. Cancer cell imaging and

photothermal therapy in the near-infrared region by using gold nanorods. J. Am.Chem. Soc., 128:2115, 2006.

[147] IV K. C. L. Black. Biomimetic Strategies Employed in the Formation of Biotargeted MetalNanoparticles for Optical Imaging and Photothermal Therapy. PhD thesis, Northwest-ern University, 2012.

[148] T. H. Meen, J. K. Tsai, S. M. Chao, Y. C. Lin, T. C. Wu, T. Y. Chang, L. W. Ji, W. Water,W. R. Chen, I. T. Tang, and C. J. Huang. Surface plasma resonant effect of goldnanoparticles on the photoelectrodes of dye-sensitized solar cells. Nanoscale Res.Lett., 8:450, 2013.

[149] S. Pillai, K. R. Catchpole, T. Trupke, and M. A. Green. Surface plasmon enhancedsilicon solar cells. J. Appl. Phys., 101:093105, 2007.

[150] D. Derkacs, S. H. Lim, P. Matheu, W. Mar, and E. T. Yu. Improved performanceof amorphous silicon solar cells via scattering from surface plasmon polaritons innearby metallic nanoparticles. Appl. Phys. Lett., 89:093103, 2006.

[151] V. J. Sorger, R. F. Oulton, and R. M. Ma andX. Zhang. Toward integrated plasmoniccircuits. MRS Bulletin, 37:728, 2012.

[152] J. A. Dionne and H. A. Atwater. Plasmonics: Metal-worthy method and materialsin nanophotonics. MRS Bulletin, 37:717, 2012.

[153] N. J. Halas and M. Moskovits. Surface-enhanced raman spectroscopy: Substratesand materials for research and applications. MRS Bulletin, 38:607, 2013.

[154] E. Smith and G. Dent. Modern Raman Spectroscopy: A Practical Approach. John Wileyand Sons, 2005.

[155] B. Liedberg, C. Nylander, and I. Lundstrom. Biosensing with surface plasmonresonance - how it all started. Biosens. Bioelectron., 10:i, 1995.

[156] E. Petryayeva and U. J. Krull. Localized surface plasmon resonance: Nanostruc-tures, bioassays and biosensing - a review. Anal. Chim. Acta, 706:8, 2011.

[157] E. Della Gaspera, M. Guglielmi, S. Agnoli, G. Granozzi, M. L. Post, V. Bello, G. Mat-tei, and A. Martucci. Au nanoparticles in nanocrystalline tio2-nio films for spr-based, selective h2s gas sensing. Chem. Mater., 22:3407, 2010.

[158] N. Liu, M. L. Tang, M. Hentschel, H. Giessen, and A. P. Alivisatos. Nanoantenna-enhanced gas sensing in a single tailored nanofocus. Nat. Mater., 10:631, 2011.

[159] J. M. Bingham, J. N. Anker, L. E. Kreno, and R. P. van Duyne. Gas sensing withhigh-resolution localized surface plasmon resonance spectroscopy. J. Am. Chem.Soc., 132:17358, 2010.

[160] J. B. Wright, K. N. Cicotte, G. Subramania, S. M. Dirk, and I. Brener. Chemose-lective gas sensors based on plasmonic nanohole arrays. Opt. Mater. Exp., 2:1655,2012.

[161] M. A. Noginov, G. Zhu, A. M. Belgrave, R. Bakker, V. M. Shalaev, E. E. Narimanov,S. Stout, E. Herz, T. Suteewong, and U. Wiesner. Demonstration of a spaser-based

Bibliogrphy 117

nanolaser. Nature, 460:1110, 2009.[162] M. I. Stockman. The spaser as a nanoscale quantum generator and ultrafast am-

plifier. J. Opt., 12:024004, 2010.[163] G. Pellegrini, G. Mattei, and P. Mazzoldi. Nanoantenna arrays for large-area emis-

sion enhancement. J. Phys. Chem. C, 115:24662, 2011.[164] A. Nevet, N. Berkovitch, A. Hayat, P. Ginzburg, S. Ginzach, O. Sorias, and

M. Orenstein. Plasmonic nanoantennas for broad-band enhancement of two-photon emission from semiconductors. Nano Lett., 10:1848, 2010.

[165] A. G. Curto, G. Volpe, T. H. Timiniau, M. P. Kreuzer, R. Quidant, and N. F. vanHulst. Unidirectional emission of a quantum dot coupled to a nanoantenna. Sci-ence, 329:930, 2010.

[166] S. A. Maier. Plasmonics: Fundamentals and Applications. Springer, 2007.[167] P. B. Johnson and R. W. Christy. Optical constants of the noble metals. Phys. Rev.

B, 6:4370, 1972.[168] S. K. Ghosh and T. Pal. Interparticle coupling effect on the surface plasmon res-

onance of gold nanoparticles: From theory to applications. Chem. Rev., 107:4797,2007.

[169] F. Bassani and G. P. Parravicini. Electron States and Optical Transitions in Solids.Pergamon Press, 1975.

[170] T. W. H. Oates and A. Mucklich. Evolution of plasmon resonances during plasmadeposition of silver nanoparticles. Nanotechnology, 16:2606, 2005.

[171] H. Ehrenreich and H. R. Philipp. Optical properties of ag and cu. Phys. Rev.,128:1622, 1962.

[172] U. Kreibig and C. v. Fragstein. The limitation of electron mean free path in smallsilver particles. Zeitschrift fuer Physik, 224:307, 1969.

[173] U. Kreibig. Kramers kronig analysis of the optical properties of small silver parti-cles. Zeitschrift fuer Physik, 234:307, 1970.

[174] U Kreibig and L. Genzel. Optical absorption of small metallic particles. Surf. Sci.,156:678, 1985.

[175] E. D. Palik. Handbook of optical constants of solids. Academic Press, 1997.[176] G. Mie. Beitrage zur optik truber medien, speziell kolloidaler metallosungen. An-

nalen der Physik, 3:25, 1908.[177] V. Bonacic-Koutecky, P. Fantucci, and J. Koutecky. Quantum chemistry of small

clusters of elements of groups ia, ib, and iia: fundamental concepts, predictions,and interpretation of experiments. Chem. Rev., 91:1035, 1991.

[178] K. P. Charle, F. Frank, and W. Schulze. The optical properties of silver microcrys-tallites in dependence on size and the influence of the matrix environment. Ber.Bunsenges. Phys. Chem., 88:350, 1984.

[179] J. M. Gerardy and M. Ausloos. Absorption spectrum of clusters of spheres from thegeneral solution of maxwell’s equations. iv. proximity, bulk, surface, and shadoweffects (in binary clusters). Phys. Rev. B, 27:6446, 1983.

[180] T. Hartling, Y. Alaverdyan, A. Hille, M. T. Wenzel, M. Kall, and L. M. Eng. Opti-cally controlled interparticle distance tuning and welding of single gold nanopar-ticle pairs by photochemical metal deposition. Opt. Express, 16:12362, 2008.

[181] J. C. Maxwell Garnett. Colours in metal glasses and in metallic films. Phil. Trans.R. Soc. Lond. A, 203:385, 1904.

[182] L. L. Zhao, K. L. Kelly, and G. C. Shatz. The extinction spectra of silver nanoparticlearrays: Influence of array structure on plasmon resonance wavelength and width.

118 Bibliography

J. Phys. Chem. B, 107:7343, 2003.[183] Sylgard 184 silicone elastomer datasheet. Technical report, Dow Corning, 2013.[184] A. Jacab, C. Rosman, Y. Khalavka, J. Becker, A. Tugler, U. Hohenester, and

C. Sonnichsen. Highly sensitive plasmonic silver nanorods. ACS Nano, 5:6880,2011.

[185] P. Winsemius. Temperature dependence of the optical properties of Au and Ag. PhDthesis, University of Leiden, 1972.

[186] W. Cai, H. Zhong, and L. Zhang. Optical measurements of oxidation behavior ofsilver nanometer particle within pores of silica host. J. Appl. Phys., 83:1705, 1998.

[187] H. Shiku, T. Saito, C. C. Wu, T. Yasukawa, M. Yokoo, H. Abe, T. Matsue, and H. Ya-mada. Oxygen permeability of surface-modified poly(dimethylsiloxane) charac-terized by scanning electrochemical microscopy. Chem. Lett., 35:234, 2006.

[188] H. Over and A. P. Seitsonen. Oxidation of metal surfaces. Science, 297:2003, 2002.[189] C. I. Carlisle, T. Fujimoto, W. S. Sim, and D. A. King. Atomic imaging of the tran-

sition between oxygen chemisorption and oxide growth on ag{111}. Surf. Sci.,470:15, 2000.

[190] V. Myroshnychenko, J. Rodrıguez-Fernandez, I. Pastoriza-Santos, A. M. Funston,C. Novo, P. Mulvaney, L. M. Liz-Marzan, and F. J. Garcıa de Abajo. Modelling theoptical response of gold nanoparticles. Chem. Soc. Rev., 37:1792, 2008.

[191] K. Chatterjee, S. Banerjee, and D. Chakravorty. Plasmon resonance shifts in oxide-coated silver nanoparticles. Phys. Rev. B, 66:085421, 2002.

[192] A. Kuzma, M. Weis, S. Flickyngerova, J. Jakabovic, A. Satka, E. Dobrocka J. Chlpik,J. Cirak, M. Donoval, P. Telek, F. Uherek, and D. Donoval. Influence of surfaceoxidation on plasmon resonance in monolayer of gold and silver nanoparticles. J.Appl. Phys., 112:103531, 2012.

[193] W. Rechberger, A. Hohenau, A. Leitner, J. R. Krenn, B. Lamprecht, and F. R.Aussenegg. Optical properties of two interacting gold nanoparticles. Opt. Com-mun., 220:137, 2003.

[194] D. J. Bergman. The dielectric constant of a composite material - a problem in clas-sical physics. Phys. Rep., 9:377, 1978.

[195] B. U. Felderhof and R. B. Jones. Effective dielectric constant of dilute suspensionsof spheres. Phys. Rev. B, 39:5669, 1989.

[196] D. J. Campbell, K. J. Beckman, C. E. Calderon, P. W. Doolan, R. M. Ottosen, A. B. El-lis, and G. C. Lisensky. Replication and compression of bulk and surface structureswith polydimethylsiloxane elastomer. J. Chem. Ed., 75:537, 1999.

[197] A. C. C. Esteves, J. Brokken-Zijp, J. Laven, H. P. Huinink, N. J. W. Reuvers, M. P.Van, and G. de With. Influence of cross-linker concentration on the cross-linkingof pdms and the network structures formed. Polymer, 50:3955, 2009.

[198] B. Prathab, V. Subramanian, and T. M. Aminabhavi. Molecular dynamics simula-tions to investigate polymer-polymer and polymer-metal oxide interactions. Poly-mer, 48:409, 2007.

[199] M. Tsige, T. Soddemann, S. B. Rempe, G. S. Grest, J. D. Kress, M. O. Rob-bins, S. W. Sides, M. J. Stevens, and E. Webb III. Interactions and structure ofpoly(dimethylsiloxane) at silicon dioxide surfaces: Electronic structure and molec-ular dynamics studies. J. Chem. Phys., 118:5132, 2003.

[200] J. E. Mark, editor. Polymer Data Handbook. Oxford University Press, 1998.[201] V. Amendola, O. M. Bakr, and F. Stellacci. A study of the surface plasmon reso-

nance of silver nanoparticles by the discrete dipole approximation method: Effect

Bibliogrphy 119

of shape, size, structure, and assembly. Plasmonics, 5:85, 2010.[202] Y. Kim, J. Zhu, B. Yeom, M. D. Prima, X. Su, J. G. Kim, S. J. Yoo, C. Uher, and

N. A. Kotov. Stretchable nanoparticle conductors with self-organized conductivepathways. Nature, 500:59, 2013.

[203] R. O. Green, M. L. Eastwood, C. M. Sarture, T. G. Chrien, M. Aronsson, B. J. Chip-pendale, J. A. Faust, B. E. Pavri, C. J. Chovit, M. Solis, M. R. Olah, and O. Williams.Imaging spectroscopy and the airborne visible/infrared imaging spectrometer(aviris). Remote Sens. Environ., 65:227, 1998.

[204] F. A. Kruse and S. L. Perry. Mineral mapping using simulated worldview-3 short-wave-infrared imagery. Remote Sens., 5:2688, 2013.

List of Publications

As of 20-12-2013

Refereed publications

• C. Ghisleri, F. Borghi, L. Ravagnan, A. Podesta, C. Melis, L. Colombo and P. Milani- Patterning og gold-polydimethylsiloxane (Au-PDMS) nanocomposites by super-sonic cluster beam implantation, J. Phys. D: Appl. Phys. 47 (2014) 015301.

• C. Ghisleri, M. Siano, L. Ravagnan, M. A. C. Potenza, and P. Milani - Nanocomposite-based stretchable optics, Laser & Photon. Rev. 7 (2013) 1020.

• G. Corbelli, C. Ghisleri, M. Marelli, P. Milani and L. Ravagnan - Highly DeformableNanostructured Elastomeric Electrodes With Improving Conductivity Upon Cycli-cal Stretching, Adv. Mater. 23 (2011) 4504.

• M. Marelli, G. Divitini, C. Collini, L. Ravagnan, G. Corbelli, C. Ghisleri, A. Gi-anfelice, C. Lenardi, P. Milani and L. Lorenzelli - Flexible and biocompatible mi-croelectrode arrays fabricated by supersonic cluster beam deposition on SU-8, J.Micromech. Microeng. 21 (2011) 045013.

Conference proceedings

• G. Corbelli, C. Ghisleri, P. Milani and L. Ravagnan - Supersonic Cluster Beam Im-plantation: a novel process for biocompatible and stretchable metallization of elas-tomers, NanotecIT newsletter 177 (2012) 31.

Publications under review

• C. Ghisleri, M. A. C. Potenza, L. Ravagnan, A. Bellacicca and P. Milani - Stretchablegrating based spectrophotometer, submitted to Appl. Phys. Lett. (2013).

Publications in preparation

• C. Ghisleri, M. Ghioni, L. Ravagnan and P. Milani - Surface Plasmon Resonanceof metal-polymer nanocomposites synthesized with Supersonic Cluster Beam Im-plantation.

121

122 List of Publications

• C. Ghisleri, F. Borghi, L. Ravagnan, A. Podesta, C. Melis, L. Colombo and P. Milani- Mechanical properties of metal-polymer nanocomposites synthesized by Super-sonic Cluster Beam Implantation.

Oral contributions to conferences and workshops

• Supersonic cluster beam implantation: a novel approach for the fabrication ofhighly stretchable and patternable elastomeric electrodes. - EuroEAP, Dubendorf(CH) - 25th-26th June 2013.

• Nanocomposite based stretchable optics - 6th International Workshop on Polymer/ Metal Nanocomposites, Toulouse (FR) - 16th-18th September 2013.

• Supersonic Cluster Beam Implantation: a novel approach for producing metal-polymer nanocomposites based microcircuits for stretchable biomedical devices. -Joint Workshop COST Actions MP1005 MP0903, S. Margherita Ligure (IT) - 7th-8thOctober 2013 - Invited oral presentation.

Poster presented in conferences and workshops

• Synthesis of soft metal-polymer nanocomposites for biomedical devices - SwissNanoconvention 2012, Lausanne (CH) - 22nd-24th May 2012.

• Synthesis of soft metal-polymer nanocomposites for biomedical devices - 6th Eu-ropean Summer School of Neuroengineering Massimo Grattarola, Genova (IT) -11th-15th June 2012.

• Supersonic cluster beam implantation: a novel approach for the fabrication ofhighly stretchable and patternable elastomeric electrodes - EuroEAP 2013, Duben-dorf (CH) - 25th-26th June 2013.

Participation to Summer Schools

• 6th European Summer School of Neuroengineering Massimo Grattarola, Genova(IT) - 11th-15th June 2012.

Patents submitted

• Method for the production of functionalized elastomeric manufactured articles andmanufactured articles thus obtained - Patent number: WO2011121017, Filling date:March 30th, 2011.

• Method for the production of stretchable and deformable optical elements, andelements thus obtained - Patent number: WO2013083624, Filling date: December5th, 2012.

Acknowledgments

This work can be considered as the end of an entire and important period of my life,projecting me in the real work world. The last three years in particular have been de-manding and unforgettable as well. They gave me the possibility to considerably growboth professionally and personally. Only now I realize how many people I worked with,I knew in the last ten years, since the first year of university in the so far 2004, and I haveto acknowledge. The main acknowledgement is addressed to Prof. Paolo Milani, whomade this PhD thesis possible and greatly supported me since the bachelor thesis. Hisreproaches and discussions gave me the possibility to meditate and to grow up.

A thanks to all the staff of our group: Marco Potenza, Alessandro Podesta, PaoloPiseri, Cristina Lenardi for their helpful discussions, Claudio Piazzoni who shared withme the troubles during the work in the laboratory, and all the secretaries, in particularNunzia Tramontano, who born all the boring administrative stuffs. A great thanks toFrancesco Cavaliere and, especially, to Daniele Vigano of the Physics Department work-shop, who built all the mechanical mounts, stands and sample holders used throughoutthe entire work. A special thanks to the PhD students and post-docs who helped meto relieve the heavy working days. First of all the new mom Francesca Borghi, withwho I sharedthe tiny desk for the last two years, Massimiliano Galluzzi and TommasoSantaniello who shared the efforts in the writing the PhD thesis. In these years theybecame excellent friends other than colleagues. Simone Bovio and Marco Indrieri firstwecolmed and introduced me in the community and in this great laboratory during thebachelor thesis: thank you. And at the end a list of all the people I had the privilege towork with (sorry if someone is missing): Luca Puricelli, Luca Bettini, Flavio Della Foglia,Michele Devetta, Carsten Schulte, Yunsong Yan, Gero Bongiorno, Pasquale Scopelliti,Giorgio Bardizza, Tommaso Mazza. A particular acknowledgement for Andrea Bellaci-cca, arrived in the last half of this work, from who I learnt a lot and who became a goodcolleague and, mainly, a confidant and friend. All the best for your next three years!

A mention to all the students who made their bachelor or master theses on stretchableelectronics and optics, essential for the results of this PhD work: Mattia Marelli, GiuliaPacchioni, Marco Turolla Turati, Marco Milesi, Andrea Barbaglia, Stefano Villa, GiacomoVigano, Mirko Siano, Alberto Annoni, Matteo Ghioni, Federico Motti.

WISE srl has been an important step of my life in the last years: it allowed me toaccess the challenging and (for me) new world of entrepreneurship and to open newways for my future. A huge thanks to Luca Ravagnan, CEO, co-founder, mentor andfirst person that introduced me to the world of metal-polymer nanocomposites manyyears ago, fundamental for the results discussed in this PhD work. I also want to gladly

123

124 Acknowledgments

thanks Marta Ferri, Sandro Ferrari, Giulia Salzano and Alessandro Antonini for the hardwork they daily perform for the success of our company.

A personal acknowledgement to my best ”out-of-the-work” friends: Paolo Ferrariabove all for sharing the bike adventures, Roberto Cuozzo and Davide Iannone, who itis always a pleasure to share a pizza and a have a walk in Bergamo with. I also remindan important person who was intermittently next to me in the last six years and I feel Imust acknowledge for her patience and comprehension.

Last but not the least a special, huge and unique thanks to my parents. There are noparticular motivations, they supported me from all the points of view and gave me thestrength to continue even in the most difficult moments. I apologize if sometimes (actu-ally many times) I did not behave respectfully or as expected because of the wearinessor stress after a heavy working day.

I keep my last acknowledgement for the person to which I am most grateful bothprofessionally and personally who began with me the work on PDMS metallization,suggested and oriented me during my master thesis and shared with me the successesand the difficulties of the first stages of WISE. You became an excellent friend and I reallymiss you a lot. Thank you Gabriele.

Date post:	19-Aug-2019
Category:	Documents
Upload:	buidung
View:	217 times
Download:	0 times

Fabrication and characterization of nanocomposite-based ... · mirrors [6,7,8], reﬂective thin...

Documents